why did electricity prices fall in england and wales?
TRANSCRIPT
OXFORD INSTITUTE ENERGY STUDIES
= FOR I
Why Did Electricity Prices Fall in England and
Wales?
Market Mechanism or Market Structure?
John Bower
Oxford Institute for Energy Studies
EL 02
September 2002
The contents of this paper are the author’s sole responsibility. They do not necessarily represent the views of the Oxford
Institute for Energy Studies or any of its Members
Copyright 0 2002
Oxford Institute for Energy Studies (Registered Charity, No. 286084)
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without prior permission of the Oxford Institute for Energy Studies.
This publication is sold subject to the condition that it shall not, by way of trade or otherwise, be lent, resold, hired out, or otherwise circulated without the publisher’s consent in any form or binding or cover other than that in which it is published and without similar condition including this condition being imposed on the subsequent purchaser.
ISBN 1 901 795 22 5
ABSTRACT
The objective of this paper is to measure the price impacts of the major regulatory interventions in the England & Wales wholesale electricity market from 1 April 1990 to 31 March 2002. More particularly, to establish whether falling prices during 1999-2002 were caused by the regulator (‘Ofgem 7 forcing the dominant duopoly of fossil fuel generators to divest coal-fired plant, or because the Pool market mechanism was replaced with New Electricity Trading Arrangements (‘NETA ’). Backward regression analysis of functionally equivalent elements of day-ahead prices under the Pool and NETA shows that changing the trading arrangements produced no statistically significant response except for the removal of Capacity Payments. In contrast, coal plant divestment, increased imports of foreign coal, and overcapacity caused by the ‘dash for gas’, caused the majority of the price fall. The direct system development costs, associated management costs, and increased operational risk of implementing NETA were therefore unnecessary because industry restructuring had already brought about a significant price reduction by March 2001. A number of fossil fuel plants, as well as the nuclear plant operators are in financial distress as a result offallingprices but as NETA had no impact on prices there is no need for further fundamental NETA reform. As in any other bulk commodity industry, prices can only rise after some firms have gone bankrupt and their capacity either sold offor closed down for good. In this case, that means the nuclear industry should not be rescued by the UK Government but allowed to fail and their assets sold to the highest bidder. Extending NETA to Scotland, as the British Electricity Trading and Transmission Arrangements (BETTA) will not benefit consumers without also breaking the Scottish generation duopoly. To achieve this, generators in England & Wales must be allowed to enter the Scottish market via an open access agreement covering the entire capacity of the upgraded Scotland-England interconnector.
1. INTRODUCTION
The England & Wales Electricity Pool (‘the Pool’) began trading on 1 April 1990 and was the
centrepiece of UK electricity market deregulation and price liberalisation. As one of the first
examples of a competitive wholesale electricity market anywhere in the world’ it was copied,
almost in entirety in some cases, by a number of other countries seeking to reform their
electricity industry.
Operationally the Pool was a mandatory uniform price auction, repeated on a daily basis, into
which generators submitted price-quantity bids to provide bulk wholesale supplies of
electricity in each half-hour of the next day (‘day-ahead,). Economic theory had predicted
that Bertrand price competition would quickly cause Pool prices to fall to short-run marginal
generation costs, however, as Figure 1 shows, mean annual Pool prices rose by 40%
I would like to thank Charles Henderson, Alex Henney, and many others in the UK electricity industry who prefer to remain anonymous, for their helpful comments on earlier drafts of this paper. All remaining errors and omissions are mine.
Various Pool-based wholesale market forms have operated in the Scandinavian countries since the early 1970s, and in Chile since 1980.
1
(nominal) in the first four years of its operation, and remained well above marginal
generation costs up to and including 2000/01.
Figure 1 : England & Wales arithmetic mean annual electricity prices and costs 1990-2002
f40.00 1
f35.00
f30.00
- f25.00
f 9 fZO.00
8 ‘C
f15.00
f l O . O O
f5.00
f O . O O 1990191 1991192 1992193 1993194 1994195 1995196 1996197 1997/98 1998199 1999100 2000101 2001102
E Z Z J Retail Price (L. Industry)
--.-Coal Marginal Cost - ’0 - CCGT Marginal Cost -A-Oil Marginal Cost
0 NETA Total Price -Pool PPP 0 Pool SMP
c.-;”IUKPX RPD
Retail Price = DTl Quarterly Energy Prices, Table 3.1.2 Prices of fuels purchased by manufacturing industry
(www.dti.gov.uk/energy/inform/energy-prices/) with quarterly price for Large consumers in UMWh calculated from
formula p/kWh * I O then averaged over Q2, Q3, Q4 plus Q1
Marginal Cost = DTI Quarterly Energy Prices, Table 3.2.1 Average prices of fuels purchased by the major UK power
producers (www.dti.gov.uk/energy/inform/energy-prices/) with quarterly marginal costs in f/MWh calculated from
formula p/kWh *IO / Thermal Efficiency (coal 33%, CCGT 45%, oil 30%) then averaged over Q2, Q3, Q4, plus QI.
Pool Price = Statistical Digest (www.elecpool.com) with prices in f/MWh averaged over Month 4-12 plus Month 1-3
NETA Total Price = UKPX RPD (www.ukpx.co.uk) plus Balancing Mechanism Costs (from anonymous
correspondent) with prices in UMWh averaged over Month 4-12 plus Month 1-3
The assumption that competition in the wholesale market would be sufficient to hold retail
prices paid by consumers at or below pre-privatisation levels proved false. As a result, the
Office of Gas and Electricity Markets2 (Ofgem), which regulates the UK electricity and gas
markets, was only able to deliver promised reductions in retail electricity prices to
householdcommercial consumers by imposing real terms reductions in the operating margins
of the transmission, distribution, and supply sectors where formal price controls were still in
place. For large industrial consumers, many of whom had enjoyed subsidised tariffs prior to
‘ Ofgem was created by merging the Office of Electricity Regulation (Offer) and the Office for Gas Regulation (Ofgas) in 1999 but for simplicity Ofgem will be used throughout.
2
price liberalisation, rising Pool prices had a more immediate impact because price controls
did not apply to electricity supplied to sites with peak demand above 1 Megawatt (MW) after
1990/9 1.
Throughout the 1990s, consumer groups representing household users, and trade bodies
representing industrial users, lobbied the UK government and Ofgem calling for action to be
taken to curtail persistently high Pool prices and hence bring about a reduction in retail
prices. Eventually, under pressure from the Department of Trade and Industry (DTI),3 who
were also increasingly concerned that high prices were encouraging excessive entry by new
gas-fired Combined Cycle Gas Turbine (CCGT) capacity to the detriment of the UK deep-
mined coal industry, Ofgem began a major Review of Electricity Trading Arrangements
in November 1997. Its interim conclusion published in May 1998 produced the
“Common Model” which proposed replacing the Pool with a bilateral market similar to a
traditional commodity market. The final report in July 1998 confirmed that the Pool would be
replaced with a bilateral wholesale market mechanism called the New Electricity Trading
Arrangements (NETA). The DTI formally announced in October 1998, that it had accepted
the Ofgem proposals.
The operational details of how NETA would function on a day-to-day basis were developed
over the next year with a start date for NETA5 of Autumn 2000. The fine detail of NETA
systems functionality continued to evolve over the next two years and difficulties in
developing and testing new software, as well as a further decision to delay the ‘Go Live’ date
beyond the Millennium, meant NETA did not begin trading until 27 March 2001.
In total, the RETA consultation process lasted a year, and the subsequent systems
development, implementation and testing for NETA took a further 30 months to complete.
During this long interregnum, Pool prices began to fall towards the end of 1999/00 and
continued to do so thereafter. Although retail prices to end-users did not fall as quickly as
wholesale prices, by the time NETA had completed its first year of operation in 2001/02 large
References to DTI generally mean the Secretary of State for Energy who was primarily responsible for the majority of regulatory initiatives relating to electricity within the department.
The key RETA documents are listed in the References section of this paper under “Ofgem” and can be downloaded from: www.ofgem.gov.uk/elarch/aback.htm, further subsidiary documents, including inquiry terms of reference, conference presentations, and third party submissions to the inquiry are available at: www.of~em.eov.ukipublic/adownloads.hlm#retabm.
The key NETA documents are listed in the References section of this paper under Ofgem and can be downloaded at: www.ofgem.aov.uklelarch/anetadocs.htm. Many more technical papers relating to the systems development can be downloaded at: http://www.ofaem.aov.uk/elarch/reta contents.htm.
3
industrial consumers were paying 15% less in nominal terms, and 61% less in real terms,‘
than they had paid in 1990/91. Such was the magnitude of the price fall that the smallest
(Fifoots, 363 MW), and largest (Drax, 4000MW), coal-fired plants both owned by AES were
in actual or technical default on their debt by the end 2001/02. BNFL, the govemment-owned
firm that had retained the UK’s oldest nuclear plants after the remainder of the industry had
been privatised, announced it would bring forward its closure programme. The privately-
owned British Energy, operating the UK’s fleet of more modern nuclear plant, announced its
revenues would not be sufficient to cover its fixed costs in 2001/02 and subsequently was
given an emergency loan by the DTI to allow it to remain in business until December 2002
until its debt could be restructured. Meanwhile Eastern Energy (TXU) and International
Power announced they would each respectively mothball coal and CCGT plant from April
2002 onwards though these were reversed once prices rose partly as a result of nuclear plant
outages, and news of British Energy’s financial difficulties emerged during summer 2002.’
Viewed in isolation, NETA and the fall in wholesale prices appear to be connected as they
both occurred over a roughly similar time period. However, NETA was only the final act in a
long series of regulatory interventions that began before the Pool commenced trading. All of
these could have had a potential impact on wholesale prices, up or down, both before and
after NETA was introduced. The objective of this paper is to measure the price impact of the
major regulatory interventions in the England & Wales wholesale electricity market between
1 April 1990 and 31 March 2002. In particular to examine the relative contribution of the
following factors to falling prices in the period 1999-2002:
i.
zz.
iii.
iv.
increasing plant overcapacity due to new CCGT plant commissioning;
increasing use of competitively priced imported coal;
divestment of coal-fired plant by the mid-merit duopoly generators; and
replacement of Pool with NETA.
..
In the remainder of this section, the main elements of the market mechanism underlying the
Pool, and NETA, are briefly described, and the diversity of opinion about its impact is briefly
surveyed. In the second section, the main regulatory interventions in the operation of the
Annual average prices deflated by annual average Retail Price Index (excluding Mortgage Interest Payments). ’ Edison Mission Energy also closed some pump storage units as NGC switched its contracting policy in the BM to make use of recently reinstated oil-fired capacity that had been mothballed under the Pool.
4
wholesale electricity market are discussed. Then the data, methodology, and results from a
backward regression analysis of year-on-year changes in day-ahead wholesale prices under
the Pool and NETA are described. In section four, conclusions are drawn about the regulatory
implications of the findings.
1.1. Pool and NETA Compared
Trading in the wholesale electricity market, under both the Pool and NETA, can be broken
down into three sequential phases, forward trading, day-ahead trading, on-the-day trading.
Each of these phases is briefly described below but for a more detailed analysis of the Pool
mechanism see Electricity Pool (1 994a, 1994b, and 1998).
Day-ahead Market
Essentially, the Pool day-ahead market was a series of mandatory uniform price auctions, one
for each half-hour of the day-ahead, operated by a centralised auctioneer with the objective of
producing an optimal schedule of generating plant to meet demand. NETA replaced this
process with decentralised voluntary bilateral contracting which could take place directly
between generators, suppliers, and traders with generating plant self dispatched by each
generator according to its contractual commitments.
The primary purpose of the day-ahead price setting process in the Pool was to produce a
least-cost schedule of plants that would be dispatched by the National Grid Company (NGC)
who operated the transmission system, to meet forecast demand in each half-hour of the next
day. The deadline for submission of bids by generators was 10.00 a.m. on the day before
dispatch with each generator submitting a nine part bid made up of three price-quantity pairs
plus a price that was intended to cover the fixed cost of starting the plant, a no load price, and
a must run flag that was used to indicate inflexible plant. Having received all the bids NGC
then created a supply curve by ‘stacking’ the bids low-to-high to produce an optimal
Unconstrained Schedule of plant for dispatch. The bid price at which the supply function
intersected with NGC’s forecast demand schedule for each half-hour of the next day set the
System Marginal Price (SMP).
5
After SMP had been set, NGC also calculated an additional Capacity Payment that reflected
the probability of supply failing to meet demand, the Loss of Load Probability (LOLP), and
the estimated marginal value of that unserved load, the Value of Lost Load (VOLL) as
follows:
Capacity Payment = LOLP x (VOLL - SMP)
Capacity Payments were therefore an entirely administered charge calculated from a formula
that was intended to compensate generators for retaining marginal plant on the system, which
they might have otherwise closed. To provide the appropriate signal, LOLP had been
designed to increase exponentially as demand approached total available system capacity. As
VOLL was set at &2000/MWh in 1990, with increases in subsequent years in line with Retail
Price Inflation (RPI), this meant that Capacity Payments could potentially become a
significant percentage of the total wholesale price, especially during the winter months when
demand was highest. The Capacity Payment was added to SMP during each half-hour to
produce the Pool Purchase Price (PPP) as follows:
PPP = SMP + Capacity Payment
All generators included in the Unconstrained Schedule received PPP for their output,
comprising SMP plus Capacity Payments where applicable, regardless of what bids they had
actually submitted. This ‘Pay-SMP’ approach resulted in those generators operating low cost,
but inflexible, nuclear and CCGT capacity consistently bidding zero in order to ensure their
plant(s) were called to run. For more than half of the Pool’s existence (until 1996/97) this
effectively left the setting of SMP, and therefore PPP, under the control of a duopoly of
firms, National Power and PowerGen, which owned or controlled almost all of the mid-merit’
coal-fired generation capacity.
Under NETA, day-ahead trading continues right up to Gate Closure, which is three hours
before the relevant half-hour dispatch period.’ As NGC is no longer responsible for producing
a day-ahead dispatch schedule all generators, suppliers, and traders, must notify NGC of the
8 .
day and night, or peaking plant that only run infrequently during periods of unusually high demand. Mid-merit plant usually only run during daylight hours when demand is relatively high, as opposed to baseload plant run continuously both
This was reduced to 1 hour in July 2002, which is after the period considered in this paper.
6
quantity of electricity they intend to physically inject or withdraw from the transmission
system in the relevant half-hour. The first notification is only indicative, known as Initial
Physical Notification (ITN), and must be submitted by 11 .OO a.m. on the day before dispatch.
A subsequent firm Final Physical Notification (FPN) must then be submitted by Gate
Closure. Although vertically integrated firms that operate both as generators and suppliers,
they must submit separate notifications for their generation and supply business, not an
aggregate net production or consumption figure. In other words, for operational purposes,
NGC still regards a vertically integrated generator with a supply business as if it were two
legally separate firms with each making an independent decision about the quantity of
electricity they will contract for, as well as inject or withdraw, in any given period.
To facilitate day-ahead, and longer-term, trading two rival market operators (APX and
UKPX) began operating screen-based exchange market places where bids and offers could be
anonymously posted. Despite this, most trading migrated to an informal over-the-counter
market, operated on the telephone and bulletin boards via brokers, and the liquidity and
transparency of the day-ahead market is lower under NETA than the Pool. However, the
UKPX publishes a Reference Price Day-Ahead (RPD) index that facilitates the comparison of
day-ahead prices under NETA with those in the Pool. Given its relative transparency, RPD
has been used in this paper as an indicator of the actual level of day-ahead prices under
NETA.
On-the-day Market
Once PPP had been set in the Pool day-ahead market generators were notified by around 5.00
p.m. that their plant would be required to run the following day. However, they were not
obliged to produce the quantity of electricity they had bid into the day-ahead market earlier
that same morning. Having been informed of PPP, generators were then free to redeclare the
physical availability of any and all of their plant right up to the moment of dispatch, though
they were not allowed to change the prices they bid. Sometimes, this was necessary because
of unexpected plant failures, and sometimes because the price of gas made it more profitable
to withdraw CCGT plant and sell the gas into the gas market. Withdrawing capacity in the
day-ahead market also had the effect of driving up LOLP, and as Bunn & Larsen (1992)
show, since LOLP is an extremely convex function strategically withdrawing capacity to
increase Capacity Payments would substantially increase the revenues earned by generators.
7
Patrick & Wolak (1997) note that a more high powered and difficult to detect strategy than
just bidding high prices to set a high SMP would be to bid each plant at close to marginal cost
and then declare capacity available in different periods throughout the day.
Withdrawing capacity on-the-day could also produce a price advantage for generators with a
large plant portfolio by forcing NGC to dispatch another of their plants not included in the
Unconstrained Schedule that had bid at a higher price. Given that dispatch of the final half-
hour of each day took place some 36 hours after the original day-ahead bid was submitted
this gave generators considerable freedom to adjust their bids.
As well as being uncertain about what plant would be available on-the-day, as compared to
that declared day-ahead, NGC could not be sure precisely what the status of the transmission
system would be in real-time. The occurrence of transmission constraints might also prevent
certain plant being dispatched as planned, or require other plant to be brought into production
that had not been included in the Unconstrained Schedule. Some uncertainty inevitably also
existed as to what the precise level of demand would be on-the-day as compared with the
NGC day-ahead forecast. The day-ahead schedule was therefore only a starting point for the
operation of the system on-the-day of dispatch and any departure from that schedule would
necessitate NGC incurring additional costs to meet outturn demand. This additional cost was
covered by the Uplift element added to PPP to make up the final Pool Selling Price (PSP) as
follows:
PSP = PPP + Uplift
Uplift contained three elements. Energy Uplift covered the incremental costs arising as a
result of errors in demand forecasting by NGC that either required plant included in the
Unconstrained Schedule to be stood down if they were no longer needed, plant not included
in the Unconstrained Schedule to be dispatched when demand was higher than forecast. In
addition, any plant not included in the Unconstrained Schedule, and not despatched, was
compensated for making plant available to the system by way of an Availability Payment
calculated in a similar way as Uplift but with an incentive for generators to bid as close to
SMP as possible. However, the inclusion of LOLP in the calculation potentially allowed
generators to raise Availability Payments in the same way as Uplift by withdrawing capacity:
Availability Payment = LOLP x (VOLL - max[SMP,bid price])
8
Transmission Services Uplift included the cost of redispatching plant where generators had
adjusted their schedules, as well as the costs incurred by NGC in bilateral purchases of
ancillary services such as black start, frequency response, voltage support, and spinning
reserve necessary to maintain the integrity, reliability, and quality of electricity supplied on
the transmission system. Reactive power Uplift, as the name suggests, covered the cost of
purchasing reactive power, usually contracted for by NGC on the same basis as other
ancillary services.
A central feature of NETA is that any costs imposed by departures from the production and
consumption patterns declared at Gate Closure were to be met by the parties that were
responsible for them, and not smeared across the industry as Uplift had been. If generators
produce more, or consumers consume less, than they had declared in their FPN then the
imbalance volume is sold by NGC at System Sell Price (SSP) in a real-time market operating
in each half-hour dispatch period called the Balancing Mechanism (BM). Likewise if
generators produce less than they declare, or consumers consume more, then NGC will seek
to eliminate the imbalance by buying additional supplies at the System Buy Price (SBP).
Imbalance costs are therefore functionally equivalent to Energy Uplift plus that part of the
Transmission Services Uplift cost that related to generators rescheduling plant. The
remainder of Uplift is functionally equivalent to the Balancing System Use of System
(BSUoS) charge. Unscheduled Availability Payments that were part of Uplift are no longer
paid for the same reason as applied to Capacity Payments. Any under or over collection of
revenue through the BM is redistributed via the Residual Cashflow Reallocation Cashflow
(RCRC) element on a pro-rata basis according to the volume of electricity physically
generated or consumed by each firm. This was substantially positive in the first few months
of NETA’s operation but gradually became negative in later months as generators
deliberately held spare generation capacity off the market ready to run if any plant declared in
the FPN failed, while suppliers deliberately overcontracted in the forward market in excess of
their expected demand. In both cases these strategies were adopted in order to avoid paying
SBP. The total cost of the BM varied from firm to firm but is calculated as follows:
BM Cost = Imbalance Costs + BSUoS - RCRC
9
A notional day-ahead price paid by a supplier under NETA that is functionally equivalent to
PSP under the Pool can be calculated as follows:
NETA Day Ahead Price = RPD + BM Cost
Forward Market
In practice generators, suppliers and consumers under both the Pool and NETA did not rely
on the day-ahead market to determine the price they paid or received. Instead, they hedged
most (>go%) of their output and consumption via forward contracts signed months or often
years ahead of dispatch. Despite the volume of electricity that was covered by contracts,
forward markets were highly illiquid and largely limited to semi-annual contracting rounds,
held in April and October. Forward market liquidity only improved as the number of
consumers eligible to choose their own supplier gradually increased, the element of cost pass
through reduced, and the ownership of generation fragmented.
In both the Pool and NETA, forward trading occurred on a bilateral basis through private
over-the-counter (OTC) contracts. The main difference being that under the Pool forward
contracting was restricted to trading in contracts for difference (CFDs) and exchange traded
forward agreements (EFAs) that were not for physical delivery but settled in cash by
reference to PPP on the relevant delivery day(s). All physical delivery took place through the
Pool and contracted plant usually bid zero prices to ensure they were included in the
Unconstrained Schedule. Under NETA, almost all forward contracts are for physical delivery
rather than financial settlement. The bilateral nature of the NETA day-ahead market, with so-
called ‘what-you-bid-is-what-you-get’ (WYBIWYG or Pay Bid) pricing, meant that baseload
generators could no longer bid zero and allow marginal generators to set the price as they had
in the Pool.
1.2. Divided Opinions
Opinion had always been divided over whether day-ahead prices had remained high because
of weaknesses in the complex market mechanism that underlay the Pool or because of the
exercise of market power by the dominant duopoly of National Power and PowerGen.
Initially, Ofgem had been unconvinced that replacing the Pool trading mechanism with a
10
mechanism where each generator got paid their own bid (Pay-Bid), rather than a uniform
marginal price (Pay SMP), would have a beneficial effect. Indeed, an Ofgem consultation on
trading outside the pool in 1994 concluded that:
In sum, paying generators their bid prices would represent a major change which seems likely to have disadvantages in terms of increasing risks, particularly to smaller generators, without a strong likellhood that prices will be lower. In the longer term it could lead to higher prices. (Ofgem July 1994)
However, by the time that the conclusions from RETA had been published in 1998 Ofgem’s
earlier misgivings about replacing the Pool appear to have been overcome. By the beginning
of 1999, considerable momentum had built up to replace the Pool, with a bilateral market
mechanism that was supposed to be closer to the operation of a traditional ‘commodity
market’. However, there was little empirical or theoretical evidence to support Ofgem’s view
that NETA would reduce wholesale prices. Despite this, an Ofgem press release on 27
January 1999” began with the headline “Pool prices must come down”. This was followed,
one week later, by another entitled “Reform of Electricity Trading Arrangements is a
Priority”. It is therefore clear that Ofgem was increasingly of the view that there was a strong
link between the trading arrangements, under the Pool, and the exercise of market power by
generators. This had inexorably lead to the conclusion that reforming the way that the market
operated was the key to lower prices. Indeed, two years before NETA was actually
implemented, Ofgem stated categorically that it believed it would result in prices some 13%
lower than had previously prevailed (FT 1999).
Outside of Ofgem, doubts still remained about whether NETA was really necessary, or even
if it would be effective in lowering prices, and some observers argued that reforming the
industry structure was more important. For example, Green (1999) suggested that the
problems of market power in England & Wales were really due to industry structure and not
Pool trading arrangements that he believed sent the correct signals to generators. He went on
to point out that under certain circumstances prices might be higher under NETA if Pool
trading were retained. Bower & Bunn (2000) also suggested that generator market power
would be enhanced under NETA, and that simpler reforms to the Pool, and a reduction in
industry concentration in the generation sector, would be just as effective in lowering prices.
10
and can be downloaded at www.of~eni.eov.uk/public/~ress releases.11 trn or hthi://wu/w.ofeem.eov.uk/public/pr.essfraanie.htm Ofgem press releases commenting on the effectiveness of RETA/NETA are listed in the References section of this paper under “Ofgem”
11
However, Currie (2002) argued strongly that NETA had already and would continue to bring
about price falls, citing the fact that forward prices had risen in September 2000 when it was
announced that the introduction of NETA was to be temporarily delayed until March 2001.
He rejected the suggestion that gradual plant divestment had brought about the price falls, not
changes to the trading arrangements, and suggested that if the Pool had remained in place
changes to the industry structure would have made little difference to the “scope for
collusion” by the mid-merit duopoly. The Ofgem Director General endorsed Currie’s
conclusions, and in a letter to the DTI (McCarthy 2001) went further by claiming that:
i. large industrial customers reported a fall in contract prices of 25 per cent over three years
in anticipation of NETA, with a further 10 per cent since NETA’s introduction;
in the over-the-counter (OTC) market, in the first three months of NETA prices fell by 6
per cent for base load, and 21 per cent for peak load; and
day-ahead base load prices had fallen by 24 per cent, on a weighted average basis,
compared with the year before.
ii.
iii.
On the first day of NETA trading, the DTI issued a press release also suggesting that NETA
had been responsible for reducing prices before it had been implemented:
The previous Pool arrangement was deeply flawed - it was effectively a means of generators setting a wholesale price which suppliers and large consumers had little choice but to accept. It was no better than a generators’ club. In contrast, NETA is a genuine market in which, for the first time, generators have to seek out customers, giving the electricity suppliers and large customers real choice.. . (DTI 2001)
Ofgem supported this conclusion with a press release in March 2001 claiming that “these
reductions have been driven by the impending introduction of NETA” and reinforced it at the
end of the first year of NETA trading in April 2002, with another press release again
implying that NETA had caused wholesale prices to fall, well before it had been
implemented, by as much as 40% since mid 1998. Both Ofgem and the DTI clearly believed
that the mere announcement that NETA would replace the Pool had been sufficient to
influence market expectations and that market participants had responded rationally by
reducing forward prices in anticipation of competition increasing under NETA. Not only that,
once NETA had begun trading those price reductions seen in the forward market began to be
reflected in prompt day-ahead prices as well. Apparently convinced of the efficacy of NETA
in promoting competition, Ofgem subsequently announced that it would be hrther enhanced
12
to include the trading of transmission rights and extended north of the border to include
Scotland by 2004 in the form of the British Electricity Trading and Transmission
Arrangements (BETTA)?
The industry remained less than convinced that NETA had had any real impact. As Maclaine
(2001a) pointed out, baseload forward prices being traded during February 2001 for delivery
in the month directly after the introduction of NETA (April 2001) were only 51.00-1.50
higher than for forward prices being traded for delivery in the month directly preceding the
introduction of NETA (March 2001). Given that generators already knew that they would
have to pay an extra &l/MWh of transmission charges under NETA, as compared to the Pool,
this suggests that market participants expected there to be a rise in post-NETA day-ahead
prices to compensate for the additional cost of transmission and the potential new risks of
trading in the BM. However, there is no indication from these forward prices that market
participants expected there to be a fall in day-ahead prices due to increased competition once
NETA began trading. McClaine went on to suggest that fragmentation of ownership in mid-
merit coal-fired generation capacity, the cancellation of the Eastern leasing arrangements on
divested plant, and the increasing amounts of CCGT plant coming on stream may all have
impacted prices well before NETA was implemented. He concluded that: “there is no clear
proof that the new electricity trading arrangements will deliver lower prices” and that it might
introduce a number of disadvantages. In a later article (Maclaine 2001b), he noted that over
the first month of NETA trading, between 27 March 2001 and end of April 2001, forward
prices had risen by 9%.
The industry consensus as to why this had occurred was that prices in the BM had been more
volatile than expected. This was especially true for SBP which had frequently been over
5100/MWh, as compared to an average of around 518/MWh for day-ahead RPD, and
therefore the incremental risk of imbalance cash out prices was being passed on by generators
along with the additional transmission charge. Indeed generators with very flexible plant
were initially able to take advantage of the volatility by bidding their plant into the BM at
very high prices though changes to NGC purchasing strategy in later months ameliorated this
to some extent. By the time that the October 2001 contracting round had been concluded it
11 Key documents relating to interim administrative wholesale pricing arrangements in Scotland after implementation of NETA, as well as
plans to include transmission pricing in NETA, and eventually extend it to Scotland, are listed in the References section of this paper under “Ofgem” and can be downloaded at: www.ofgem.gov.uk/proiects/bctta index.htm.
13
had become clear that NETA had not brought any significant benefit to consumers in terms of
lower prices and overall prices had risen by a couple of percentage points since the previous
year (Maclaine 2002).
In a preliminary analysis of the first year of NETA, Newbery (2002) concluded that it had
been a "costly mistake that others should avoid" and by the time that Ofgem produced its
formal review of the first year of NETA in July 2002 it appeared to have retreated from its
earlier firmly held view that NETA was entirely, or mainly, responsible for the fall in prices
between 1999/00 and 2001/02. It now appeared to accept that other factors could have played
a part in the reduction in prices:
Anticipation of NETA appears to have been a clear influence on the downward trend in forward contract prices from 1998. However, other factors, including further developments in the generation market, and a stable plant margin are also likely to have played a part. Over this period fuel prices have remained broadly constant (coal) or increased (gas) in real terms. Ofgem (July 2002).
A year after the introduction of NETA it was therefore still an open question, as to what the
relative impact of replacing the Pool with NETA, industry restructuring, and the long series
of additional regulatory interventions that occurred throughout the history of the Pool, had
been on wholesale electricity prices in England & Wales.
14
2. REGULATORY INTERVENTIONS
The UK Government in the form of the Treasury, Competition Commission,’’ DTI, and
Ofgem, intervened at various times, to modify the way that the wholesale electricity market
operated in England & Wales from 1990 onwards. As a result, the Pool never really operated
as a truly free market. In the absence of effective competition Ofgem was particularly active
in controlling wholesale, and retail, prices through a series of regulatory interventions that
had the implicit or explicit objective of placing a cap on prices. As Figure 2 shows, from
1996/97 onwards the frequency and duration of these regulatory interventions increased
dramatically as compared with the previous six years.
2.1. Coal Contracts
On 31 March 1990, all of the coal-fired and oil-fired generating plant in England & Wales
that had previously been under the control of the state-owned Central Electricity Generating
Board (CEGB) were allocated (‘vested’) to two new companies, National Power, and
PowerGen, in a 60:40 ratio.” The Regional Electricity Companies (RECS)’~ also had
matching vesting contracts imposed upon them that obliged them to buy fixed amounts of
electricity from the generators at a price that would provide a guaranteed margin to the
generators and RECs. These vesting contracts, which lasted fiom 1 April 1990 to 31 March
1993, therefore effectively subsidised the UK coal industry by passing the cost of its deep-
mined coal, which was higher than world prices, through to end users. Generators and RECs
were therefore insulated from both the impact of high input prices as well as volatility in the
Pool price.
During the term of these contracts the combined contractual volume of UK deep mined coal
consumed by UK generators was 70 million tonnes per year, in 1990/91 and 1991/92, then 65
million tonnes for 1992/93. Despite direct intervention by the DTI in the negotiations,
generators were unwilling to continue contracting for large tonnages of UK coal after 31
12
l3 For a chronology of the key events in the legislation, deregulation, privatisation, regulation of the UK electricity industry between 1988 and 2002 see Electricity companies in the United Kingdom - a briefchronology at httr,://www.clcchicitv.ore.ukiservices fr.html. 14 RECs were vertically disintegrated from generation at vesting, as was the National Grid Company (NGC) that owned and operated the transmission system. Initially each REC operated exclusively in one of 12 regions, where it had monopoly control of the distribution system and monopoly rights to supply household and commercial consumers with peak demand below 1 MW, the so-called ‘franchise market’. After 3 1 March 1993, the franchise market threshold was reduced to IOOkW, and the RECs finally lost their remaining supply monopolies in 1998/99 when domestic and small commercial consumers were allowed to choose their own supplier.
Formerly the Monopolies and Mergers Commission (MMC).
15
March 1993 because of the prospect of coal plants increasingly being displaced by new
entrant IPPs running CCGT plant. The deal that was eventually agreed saw contracted UK
coal fall to 40 million tonnes for 1993/94 and 30 million tonnes for the following four years
(Parker 2000). Thereafter, the UK coal industry essentially had to compete with imported
supplies at world market prices and much of the remaining deep-mine capacity closed as a
result. In 2000/01, total UK deep-mined coal production fell to 18 million tonnes, as
compared to 34 million tonnes of imports (DTI 1990-2002a and 1990-2002b). Newbery
(1995) argued that if a generator has sold forward contracts equal to the amount dispatched in
a given period then its income is determined by the strike price of the forward contract not by
the Pool price, and therefore it would have no incentive to raise SMP as this would not affect
its revenue. Indeed the firm could do no better than bidding at short-run marginal cost.
Vesting contracts had a pro-competitive effect as they encouraged all generators to bid their
plants as if the market were perfectly competitive. Thus vesting contracts reduced the
incentive for mid-merit generators to exercise market power. As the volume of vesting
contracts declined, the incentive for mid-merit generators to exercise market power in order
to raise Pool prices increased. Indeed, as each anniversary of vesting approached pool prices
did rise and the consensus view was that the duopoly generators used this as a signal to
buyers that forward prices in the next contracting round would be above previous levels.
2.2. Pool Price Cap
As Pool prices had consistently risen since 1990 Ofgem had little choice but to place an
explicit cap on them15. By threatening to refer the matter to the Competition Commission,
Ofgem had extracted an undertaking from National Power and PowerGen that they would bid
their plant in such a way as to ensure that mean annual demand weighted prices did not rise
above &24.00/MWh on a time-weighted and &25.50/MWh on a demand-weighted basis (in
October 1993 prices) for two years between 1 April 1994 and 3 1 March 1996. Although Pool
prices did slightly exceed the target level in 1994/95 on a demand-weighted basis, they were
a few percent below the cap on a time-weighted basis and no punitive action was taken.
However, as plant owned by National Power and PowerGen had set SMP over 90% of the
time throughout the duration of the cap, and outturn price levels had so precisely matched the
l5 Key documents relating to the regular inquiries into price levels, and interventions such as the imposition of a price cap, are listed in the References section of this paper under “Ofgem” and can be downloaded at httn://www.of~em.~ov.uk/public/adownloads.htm#pool
16
level agreed with the regulator some two years earlier, this suggests that they had exercised
an extraordinary degree of control over the level of prices.
2.3. Plant Divestment
The Pool price cap was only ever intended to be a temporary measure and integral to a wider
agreement reached by Ofgem with National Power and PowerGen that they would
respectively divest 4000 MW, and 2000 MW of capacity in the intervening period.
Eventually, all the capacity was leased, but not sold, to Eastern Electricity, a REC in mid
1996. The terms of these deals meant that Eastern paid a fixed charge, at the beginning of
each year, plus a variable charge of around &6/MWh of output, that effectively raised its
marginal production cost, and so prevented it being used to undercut the remaining National
Power, and PowerGen, plant portfolio.
It was hoped that the plant divestment, to create a third mid-merit generating firm, would
increase competition sufficiently to keep prices under control once the price cap was
removed. However, despite the divestment of plant to Eastern, arithmetic mean annual PSP
only fell by about &l/MWh over the next four years. Though prices had on average been
lower in 1997/98 than under the Pool price cap, severe price spikes in winter of that year
promoted a further round of investigations into the price setting behaviour of the duopoly
generators which carried on for most of 1998. This was quickly followed by yet another
investigation into the price spikes that occurred in January 1999. Almost as soon as this
review had finished in May 1999 further spikes occurred in July, reported on in October of
that year. Prior to the imposition of the price cap there had been a similar litany of
investigation beginning as early as 1991, and it quickly became clear that this first round of
plant divestment had only reduced mean annual prices slightly. Not only that, prices had
become significantly more volatile, especially during the winter months. Green & Newbery
(1992) had argued that five price-setting (i.e. mid-merit) generators would be sufficient to
ensure a competitive market. Green (1996) extended the work to show that although there
would be an increase in competitiveness of the market due to the partial divestiture it would
not be as large as if the duopolists had been split into several much smaller firms.
Fundamentally, the creation of a third, relatively small generator, left the market substantially
more concentrated than it would have been had five equal sized generators been created and
the empirically observed price response to the divestment is not surprising.
17
Ofgem and the DTI came under further pressure to take action and, simultaneous to the
RETA public consultation process, they entered into a joint negotiation with National Power
and PowerGen over further plant divestment. Eventually it was agreed that each firm would
be allowed to vertically reintegrate its generation businesses with the supply business of a
REC in return for further divestment of around 4000 MW of plant. As a result, PowerGen
bought the retail supply business of East Midlands Electricity, in July 1998, and National
Power bought the retail supply business of Midlands Electricity Board (MEB) in June 1999.
Subsequently, PowerGen completed the sale of their Fiddlers Ferry (1960MW) and
Ferrybridge (1956MW) plants in July 1999 to Edison Mission Energy. National Power
likewise sold its Drax plant (3870MW) in November 1999 to AES. In each case the
divestments amounted to approximately 40% of their respective coal-fired plant portfolios. In
addition, it was agreed that all of the lease arrangements on Eastern plant that had previously
been divested were to be cancelled in exchange for a one-off lump sum payment. Despite the
public pronouncements about the likely effectiveness of NETA, the OfgedDTI actions to
force further plant divestment suggest they were less than wholly confident that reforming the
electricity trading arrangements would be sufficient to reduce prices to a competitive level.
After the second mandatory round of divestment, National Power and PowerGen
management teams appear to have concluded that, in light of the prices that potential new
entrants were willing to pay for existing coal plant, and in order to avoid further regulatory
confrontation, it would be in the best interests of their shareholders to make further voluntary
plant divestments. Over the next three years PowerGen reduced its capacity share of the coal
generation sector down to 13% by divesting further plant to British Energy and the French
utility EDF. Eventually, the entire firm was taken over by E.ON a German utility in July
2002. National Power reduced its coal generation capacity share to a similar level and then
separated its now vertically integrated UK business from its international business to form
two new companies, Innogy that was subsequently taken over by the German utility RWE in
May 2002, and International Power. Eastern Electric also divested coal plant and a wave of
generation asset trading, including CCGT and oil plants, took place throughout 2000/01 to
2001/02. The result was that, by the time NETA began trading, the coal-fired generation
sector became fiagmented between eight firms. The industry structure of the mid-merit coal-
fired generation sector that prevailed during the first year of NETA was therefore starkly
different to that over the life of the Pool, as shown in Figure 3.
19
100%
80%
60%
40%
20%
Figure 3: Coal-fired generation capacity ownership structure 1990-2002
2.4. Review of Electricity Trading Arrangements
National Power
~ B E G L
minternational Power
mlnnogy
~ A E S
CaTXUlEastern
EdF
~ A E P
BEdison
0 PowerGen
DScottish Power
~ A L C A N
0%
Beginning in May 1998, RETA was launched with the stated aim of developing an entirely
new wholesale market mechanism to replace the Pool. Ofgem identified the following major
weaknesses in the Pool trading arrangements:
i.
ii.
iii.
iv.
V .
price setting in the Pool is overly complex since it required the submission of at least
nine different bid parameters for each genset with the merit order effectively produced
from a ‘black box’ optimisation program that lacked transparency;
capacity and availability payments reward generators for making plant available, not
operating it, and it has limited value as a price signal to the generation or demand side
of the market to respond to short-term changes in market conditions;
bids do not reflect costs as many baseload generators consistently bid a zero price, so-
called zero-zero bids, relying on the mid-merit generators to set SMP;
prices have risen substantially, and become increasingly volatile, since the Pool began
trading even though fossil fuel prices have fallen;
market liquidity, and the lack of publicly available price data for forward contracts,
puts consumers at a disadvantage when negotiating forward cover against Pool prices;
20
vi. the non-firm nature of the day-ahead market transfers costs and risks of plant failures
from generators to consumers, through Energy Uplift payments;
the security of electricity supply is being threatened because generators can sign
cheap ‘interruptible’ gas supply contracts, or sell gas from ‘firm’ gas supply contracts
when spot gas prices rise, without paying a penalty in the electricity market; and
the participation of the demand side in price setting is limited to a few very large
industrial consumers.
vii.
viii.
Consumer groups had generally echoed these concerns in their submissions to RETA.
Outside commentators had also made similar criticisms (Newbery 1997) and proposed a
number of alternative models that Ofgem synthesised into the final NETA proposal which
sought to address the weaknesses listed above by essentially adopting trading arrangements
mimicking those in traditional commodity markets. These included:
1.
.. 12.
iii.
iv.
V.
vi.
vii.
simplifying generators’ bids so that they are only for a given price, quantity, and
delivery period which means that generators must internalise all the associated costs
of start-up and fixed costs of operating a plant;
eliminating capacity and availability payments;
forcing all generators and consumers to compete more actively in setting market
prices by introducing Pay Bid pricing;
introducing screen-based trading, and a balancing market, to promote real-time price
transparency and encourage independent price reporting as in other commodity
futures markets;
making all bids and offers ‘firm’ which means that a generator must deliver, and a
consumer take delivery, against their contracted positions, or face the uncertain
consequences of the system operator (NGC) buying or selling in the balancing
market, on their behalf, and passing the costs back to them;
imposing the full cost and risk of interruptible gas contracts and firm gas sales, back
on the generators, and allowing the IS0 to buy or sell in a balancing market against
any under or over delivery against notified contract positions; and
giving large consumers, and suppliers, an incentive to undertake active load and price
risk management, and allow the IS0 to buy or sell against any over, or under
consumption, against their notified contract purchases.
21
2.5. Gas Moratorium
Demand-weighted average prices for coal bought by UK generators fell as volumes of low
cost imported coal took an increasing share of the UK market. However, the guaranteed
generation margin that vesting contracts provided had the unintended effect of encouraging
incumbent generators, the RECS, and new entrants to build new CCGT plant. These early
CCGT entrants were able to undercut small and medium sized coal plant, on a short-run
marginal cost basis, and also displaced more efficient large coal plant from the baseload
segment of the load duration curve. A combination of new entry by CCGT, and rising oil
prices, also meant that much of the conventional oil-fired generation capacity was closed or
mothballed under the Pool. Increasing thermal efficiency through rapidly advancing turbine
technology, falling gas prices, falling capital costs, and relatively short construction times for
CCGT lead to the 'dash for gas' that resulted in 22500 MW of new capacity being
commissioned by 2002. Figure 4 shows the capacity share of each generation technology
throughout the life of the Pool and first year of NETA. The technologies have been sorted
according to their approximate marginal production costs and this shows how coal plant was
displaced by the rapid entry of CCGT that amounted to around one third of total generation
capacity by the end of 2001/02 and approximately 40% of generation output.
Figure 4: Capacity share by fuel and generation technology 1990-2002
~ 100%
90%
80%
I UOCGT
70%
60%
50%
40%
30%
20%
10%
0%
22
By the mid 1990s, the DTI was becoming alarmed at the increasing dependence of the UK on
gas-fired generation capacity and in November 1997, coincident with launching RETA, it
imposed a moratorium on issuing of new CCGT plant construction licences. This was
designed to provide a breathing space during which energy policy could be reviewed,
especially in relation to UK coal, and a final decision to be made about whether to reform or
replace the Pool trading arrangements. When the DTI published its final report in June 1998
(DTI 1998), it concluded that the growth in CCGT capacity, and depletion of gas reserves in
the North Sea, meant that the UK was likely to become a net importer of gas by 2010.
Moreover, growth projections to 2020 suggested gas could account for 60% of total UK
primary energy consumption, and 75-90% of electricity generation fuel by 2020 (DTI 2000).
The main proposal coming out of the review was that the gas moratorium should continue
until the electricity market had been reformed, on the grounds that if no action was taken “the
country could effectively lose a coal option”. The clear intention was to allow NETA
sufficient time to reduce wholesale prices to a level that would make further CCGT
investment unattractive and preserve the remaining coal-fired generation capacity. However,
Ofgem disagreed with the Gas Moratorium policy and publicly stated that:
The policy of restricting new entry into generation is now the main obstacle to a more competitive electricity market. A continuing threat to incumbents from new entrants is of the utmost importance in preventing collusion, and in ensuring that reforms to trading arrangements are successful and deliver benefits to consumers, particularly in the form of lower prices. (Ofgem 1999)
In practice, little could be done to increase the amount of UK coal contracted for by
generators after 31 March 1998, even if all the existing coal-fired capacity remained open.
Facing intense lobbying from foreign firms who wished to enter the UK electricity market
with new CCGT plant, and without the support of Ofgem, the DTI had little choice but to
abandon its policy. Once RETA had been completed, and the DTI and Ofgem had agreed the
structure of NETA in October 1999, the gas moratorium was formally lifted in November
1999. As it did not cover plants already under construction, or that had already been licensed,
the moratorium’s impact was limited to delaying the construction of a few plants by a matter
of a year to eighteen months.
Even if the gas moratorium had not been lifted, it is likely that compensating mechanisms in
the fuel and capital markets would have rendered it unnecessary anyway. As Figure 1,
23
showed, the artificial cost advantage that CCGT enjoyed due to the coal vesting contracts had
already been eroded by 1997/98 as the quantity of imported coal increased. Short-run
marginal costs for coal and CCGT converged which meant that in the later years of the Pool
and first year of NETA coal and gas were competing head-to-head for fuel share of the
electricity market. Under NETA, the BM rewarded flexible plant that could run at short
notice and generators reinstated oil plants to take advantage of occasional very high SBP,
displacing lower cost but relatively inflexible CCGT and coal plants in the merit order.
2.6. Market Abuse Licence Clause
In the intervening period, between the completion of RETA and the introduction of NETA,
Ofgem attempted to control the wholesale price of electricity by introducing a new market
abuse licence clause (MALC) in generation licences.“ The objective was to limit the scope
for generators to exercise market power throughout the remaining life of the Pool, and for at
least the first year after the introduction of NETA. However, there was no definition of what
price level, or trading activity, would constitute market abuse. Indeed, Ofgem explicitly
stated that it was against the introduction of any form of price regulation because it was
“likely to stifle innovation, have harmful effects on competition by limiting the incentives on
new entry into the market and, additionally, have a detrimental effect on the development of
the forward contract market”. However, the proposal effectively gave Ofgem absolute
discretion to determine a maximum permissible level for wholesale market prices, and define
any trading activity as market abuse. By introducing MALC, it appears that Ofgem was
attempting to signal to generators that wholesale market prices would either be reduced to
competitive levels by competition or, if necessary by regulatory intervention.
In effect, Ofgem gave generators little option but to accept MALC and reduce Pool prices.
Only two generators refused to accept the modification to their generation licences, AES and
British Energy, and both were immediately referred to the Competition Commission.
However, the findings of the subsequent inquiry did not support Ofgem. The Competition
Commission concluded that MALC was unnecessary in the context of the plant divestment
that had taken place already, and further divestments that were likely to take place in future,
as well as new entry by CCGT plant operators. It also noted the 17% fall in real terms in Pool
l6 Key documents relating to the introduction of MALC and the Ofgem response to the Competition Commission decision are listed in the References section of this paper under “Ofgem” and can be downloaded at: htta://www.ofeem.eov.uWuublic/pubframe.htm.
24
prices that had already occurred in the five years up to 1999/2000 (Competition Commission
2001). In the particular case of AES and British Energy, the Competition Commission noted
that the behaviour of the two firms in forward contracting substantially all of their output on a
long-term basis made it virtually impossible for them to exercise market power either by
withdrawing capacity or raising the price at which they offered that capacity to the market.
In view of the Competition Commission decision, Ofgem was forced to remove MALC from
all the generator licences that had already agreed to its introduction. However, the Ofgem
Director General's comments in response to the decision reveals that he was far from certain
that the plant divestment that had already taken place, andor the introduction of NETA in a
few months time, would be sufficient to deliver lower prices without MALC:
I regard this as a substantial loss of measures needed by Ofgem to do our job. This is to the detriment of consumers whose protection is Ofgem's primary objective under the law. Ofgem will continue to look at other ways of tackling market abuse although none of these is likely to be as effective as the licence condition which we have now had to withdraw. (Ofgem 2001)
With the backing of the DTI, Ofgem attempted to reintroduce MALC by producing fresh
proposals in August 2001 but in a form that might be more acceptable to the Competition
Commission. However, as wholesale prices continued to fall throughout 2001/02, it became
increasingly untenable to argue that MALC was needed in any form and the matter was
quietly dropped.
25
3. ANALYSIS
In the first six years of the Pool’s existence, up to the end of 1995/96, regulatory
interventions fell into two distinct sequential phases. The first phase was the vesting contracts
that were essentially driven by the UK Treasury’s need to facilitate privatisation of the
electricity industry and provide support to UK coal without recourse to public hnds. The
second phase was the imposition of the Pool price cap beginning on 1 April 1994. However,
in the period from 1996/97 onwards, covering the last five years of the Pool and the first year
of NETA, Ofgem and the DTI appear to have abandoned the earlier incremental approach to
regulation in favour of a more aggressive multi-layered approach that manifested itself in a
continuous stream of overlapping interventions. Against this background, the claims and
counterclaims about the relative effectiveness of NETA in reducing wholesale prices, versus
any of the other regulatory interventions that occurred over the same period, are difficult to
assess.
3.1. Analytical Method
In the remainder of this section an attempt is made to measure precisely the significance, and
quantify the impact on day-ahead prices, of each of the major regulatory interventions
described above. To achieve this, dependent variables representing different functionally
equivalent components of the wholesale price under the Pool and NETA are regressed on a
range of explanatory variables representing each of the major regulatory interventions, as
well as external factors such as weather, and plant investment. The objective is to identify
which, if any, of these interventions were responsible for the year-on-year wholesale price
changes that were observed over the period 1990/91-2001/02. Table 1 summarises the data
sources, units, and time periods covered by each of the dependent and independent variables.
A backward linear least squares multiple regression procedure was applied to the variables in
which the initial regression run contains all the candidate independent variables. The
independent variable with the highest p-value (i.e. statistically least significant) is then
eliminated and the regression repeated. The sequential elimination of insignificant
independent variables is repeated until the p-values of all the remaining variables are equal to
or less than 0.10. In other words, the final regression model only contains independent
variables that have coefficients that are statistically significant at the 90% confidence level.
26
Wolfram (1998) analysed time series of Pool prices with a multiple regression approach to
measure the impact of regulatory interventions on Pool price mark up over short-run marginal
generation costs. From the results, she was able to conclude that generators were not taking
full advantage of their ability to exercise market power, given the inelastic demand for
electricity, but that they were modifying their behaviour in response to regulatory
interventions, and that generators were restraining prices either to deter entry or stave off
substantial regulatory action. Although the approach proposed in this paper does not attempt
to replicate these results, this earlier study does indicate that it is feasible to detect the
response of generators to regulatory interventions from an analysis of price time series and
provides support for the choice of analytical approach. The construction of the variables, their
analysis, and the results are discussed in detail throughout the remainder of this section.
3.2. Dependent Variables
The starting point for the analysis was the construction of Pool day-ahead wholesale price
time series. Monthly arithmetic mean SMP, PPP, PSP, were collected from the period 1 April
1990 to 26 March 2001 (Electricity Pool 1990-2002). A Capacity Payment time series for the
same period was calculated by subtracting SMP from PPP, and an Uplift time series
calculated by subtracting PPP from PSP. NETA wholesale price time series were calculated
for the period 1 April 2001 to March 2002.” Monthly arithmetic means for the day-ahead
Reference Price Data index (WD) was calculated from hourly data supplied by UKPX.
Capturing a representative value for the BM cost was more difficult since each firm pays (or
receives) a different amount depending on the accuracy with which it forecasts its demand,
and what contracting strategy it adopts before gate closure. As a result, a BM cost could only
be estimated for a theoretical supplier that was assumed to be producing an unbiased forecast
of its demand on a day-ahead basis, and deliberately purchasing contracts to cover 102.5% of
its forecast demand. It was further assumed that the supplier carries out the final 2.5% of
overcontracting in the day-ahead market at mean monthly RPD and that the actual outturn
demand, faced by theoretical supplier on-the-day, is symmetrically distributed between
97.5% and 102.5% of its day-ahead demand forecast. Any overcontracted supply in excess of
actual demand is subsequently sold (‘spilled’) at SSP in the BM.
Pool prices for March 2001 are calculated up to 26 March 2001 when the Pool ceased trading. NETA began trading on 27 March 2001, but April 2001 monthly average RPD is calculated from 1 April 2001.
27
I L
i
c
* Q Q a
E - i .. (U
" E
c-r a 2 3
.. In s P
The theoretical firm is therefore never exposed to SBP, but is on average purchasing 2.5% of
its day-ahead forecast demand at mean monthly RPD, in the day-ahead market, and then
selling it at SSP in the BM. The cost of imbalances is therefore equal to the difference
between RPD purchase price and the SSP sale price multiplied by the volume of over
contracting. Although this is a simplifying assumption, it captures the forward contracting
strategy that has increasingly being adopted by supply firms, and represents the costs that
would have been faced by those employing state-of-the-art forecasting methods. Daily data
for BSUoS, RCRC, and SSP were collected from Elexon" and monthly arithmetic mean time
series constructed for each over the period 1 April 2001 to 31 March 2002. Mean monthly
BM costs were then calculated from the following formula:
BSUOS + RCRC + (RPD - SSP) x 2.5%))
Raw wholesale price time series for the period 1 April 1990 to 3 1 h d c h 200 (1 44 monthly
observations) covering the entire history of the Pool and first year of NETA were constructed
by concatenating the following combinations of raw time series data described above:
i. SMP and RPD
ii. PPP and RPD
iii. PSP and RPD plus BM Cost
These series were then transformed to produce three dependent variables, respectively called
SMPRPD, PPPRPD, and PSPNETA, each consisting of a series of 132 monthly year-on-year
price changes, using the following formula:
PDifferenced = PCurrent Month - PMonth -12
A CAPACITY variable consisting of monthly year-on-year changes in Capacity Payments
was constructed by subtracting the SMPRPD series from the PPPRPD series. Likewise, an
UPLIFTBM series, consisting of monthly year-on-year changes in Uplift Payments and BM
costs, was constructed by subtracting PPPRPD from PSPNETA.
l8 Elexon is the firm that was set up to oversee clearing and settlement of trades in the BM. Summary data are available on the www.bmreaorts website, however, I am grateful for the generous assistance of an anonymous third party who made a more detailed dataset available to me in electronic form that was used in this paper.
29
3.3. Independent Variables
As for the dependent variables the process began by constructing time series consisting of
144 monthly observations stretching from 1 April 1990 to 31 March 2002.
To capture the impact of changes in ownership of generation capacity, the fuel source,
registered capacity, and ownership of every power plant connected directly to the
transmission system of England & Wales” was tracked on a monthly basis using publicly
available data published by CEGB (1989), NGC (1991-2002), Electricity Association (1991-
2001), Power UK (1994-2002), as well as press archive searches. In the absence of
contradictory data, plant closures were assumed to take place on 31 March, which is the
deadline imposed by NGC on generators to declare plant as Transmission Contracted for the
following year. Commissioning of new plant, especially CCGT is assumed to take place on
the first day of the month following that in which plant testing and final hand-over from the
contractor to generator were completed. The evolution of capacity on interconnector
transmission circuits that allow imports from France and Scotland into England & Wales was
tracked from the same sources as for plant capacity.
A Capacity Concentration Index (CCI) was then calculated on a monthly basis for Coal,
CCGT, Nuclear, and Oil plant, by summing the squared capacity share of each firm for each
plant type. A CCI for the entire industry was also calculated by summing squared shares of
total capacity owned by each firm, regardless of plant type. However, as changes in the
Industry CCI essentially tracked the change in CCI of the underlying plant technologies this
variable was not used in the final analysis in order to avoid multicollinearity between
variables. The Open Cycle Gas Turbine (OCGT) CCI was not used because these small
auxiliary plants were generally operated from the same site as coal plant and changes in
OCGT CCI were therefore highly correlated with changes in Coal CCI. The CCI for hydro,
and pump storage, plant were also calculated but not used as both were constant throughout.
Interconnector capacity from France was included in the nuclear CCI, and interconnector
capacity from Scotland was split on a pro-rata basis between hydro and coal capacity to
reflect the interconnector shares and generation capacity held by the two Scottish generators,
19 This Transmission Contracted Plant excludes the embedded generation capacity connected directly to local distribution systems, which was excluded from the obligation to submit bids into the Pool.
30
Scottish & Southern Energy and Scottish Power respectively. Any dual fuel plant on the
system was assumed to be burning coal and included in the Coal CCI.
The CCI is non-linear with a theoretical maximum of 10,000 if one firm owns all of the
capacity of a given type and zero if that capacity is spread between an infinite number of
firms. It is therefore identical in concept to the Hirschman-Herfindahl Index (HHI),
commonly used by competition authorities to measure the impact of industry concentration.
The only difference being that the sum of squared market shares in an industry, used in the
HHI, has been replaced with the sum of squared capacity shares in the CCI. At the inception
of the Pool in 1990, National Power, and PowerGen, owned all the coal-fired generation
capacity in an approximate 60:40 ratio, corresponding to a Coal CCI of around 5200. By
2001/02 the Coal CCI had fallen to around 1400 because of plant divestment. By comparison,
an industry with coal capacity split equally between 5 firms, as proposed by Green &
Newbery (1992), would have resulted in a Coal CCI of 2000. As the first CCGT plant in
existence did not become fully operational until around November 1991, the CCGT CCI is
assumed to be 10,000 before this date and thereafter until the second CCGT plant came on
stream in April 1993 when the CCGT CCI begins to fall rapidly.
Short-run marginal generation costs were calculated for Coal, CCGT, and Oil plants based on
the average cost of each fuel used in UK power stations, and the estimated thermal efficiency
for a marginal plant of each type, as reported in DTI statistics (DTI 1990-2002a and DTI
1990-2002b) using the following formula:
Marginal Generation Cost = (Fuel Price in p k w h / Thermal Efficiency in %) x 1000
Average thermal efficiencies have gradually increased for the entire fleet of UK Coal plant
from 34% to 36% and for CCGT from 43% to 50%, over the period 1990-2001 but marginal
thermal efficiency data were not reported. For simplicity, it has been assumed that they
remained constant at 33% for coal, 45% for CCGT, and 30% for oil.
Data on annual tonnages of UK deep-mined coal production were also collected from DTI
statistics (DT1 1990-2002a and DTI 1990-2002b) in
contracted volumes of UK coal sold to power generators.
an attempt to estimate annual
However, after some preliminary
31
testing it was found that contracted tonnages reported by Parker (2000), and drawn from
searches of press archives, were a relatively more powerful explanatory variable.
The impact of new entry was modelled by estimating the plant margin for each month. As
data on total system demand, and the quantity of plant available to the system after outages
were not available, a proxy was estimated from total plant capacity and annual peak demand
(NGC 1991-2002) and (EA 1991-2001) as follows:
Plant Margin Mthly = (Plant Capacity Mthly - ACS Peak yrly) / (Plant Capacity Mthly x 100)
The final quantitative variable considered was temperature. To capture its effect on prices,
mean monthly temperature data as well as deviations from the long-tern mean were
collected. Mean Heating Degree and Mean Cooling Degree parameters were also calculated
from the following formulas:
Mean Heating Degrees = Max (1 5 - Mean Monthly Temperature,O)
Mean Cooling Degrees = Max (Mean Monthly Temperature - 15,O)
After preliminary analysis, it became apparent that Mean Heating Degrees was the only
significant temperature variable, and the others were discarded. This is consistent with UK
peak electricity demand occurring in the winter, but this variable captures any increase in
electricity demand above and beyond that normal winter seasonal pattern. As UK winter
temperatures are generally higher than for similar north European latitudes, periods of
unusually cold temperatures tend to have a disproportionate effect on demand for electricity,
and hence price. Though the majority of UK space heating is from gas-fired and oil-fired
boilers, relatively inefficient electrical appliances are often used to meet incremental heating
demand above normal winter levels, especially for domestic and small commercial users.
Regulatory interventions including the Pool price cap, RETA, MALC, Gas Moratorium and
NETA are modelled as qualitative factors with dummy variables taking a value 1, for each
month that an intervention was in place, and 0 in months where it was not. For example,
NETA is represented by a dummy variable taking a value 1 in the twelve months fiom 1
April 2001 to 3 1 March 2002, and 0 in all preceding months. The period covered by the lease
32
arrangements on the plant divested to Eastern was also modelled with a dummy variable but
the concentration effect of the disposal was also separately accounted for in the Coal CCI.
As with the dependent variables, all of the raw time series described above were transformed
to construct a set of independent variables, each consisting of a time series of 132 monthly
year-on-year changes.
3.4. SMP-RPD Results
The SMPWD series was analysed by backward regression, as described above, and a
summary of the resulting coefficients and selected test statistics is set out in Table 2.20
Table 2: SMPRPD regression output
Variable Coefficient Std. Error t-statistic p-value
CAP CCGTCCI COALCCI COALCONR N ETA OlLCCl OILMC C
-1.7893 0.001 0 0.001 9
-0.0002 3.8931 0.0056
-0.3356 1.3765
0.8997 0.0003 0.0008 0.0001 1.4710 0.001 0 0.0899 0.4573
-1.9887 3.6285 2.2951
2.6466 5.6266
-2.5949
-3.7331 3.0100
0.0489 0.0004 0.0234 0.0106 0.0092 0.0000 0.0003 0.0032
R-squared 0.401 1 Mean dependent var -0.0167 Adjusted R-squared 0.3672 S.D. dependent var 4.3227 S.E. of regression 3.4385 Akaike info criterion 5.3667 Sum squared resid 1466.1075 Schwarz criterion 5.5414
1 1.861 7 Log likelihood -346.1 991 F-statistic Durbin-Watson stat 1.3775 Prob(F-statistic) 0.0000
The most important result is for the NETA variable that suggests that the introduction of
NETA had no effect on curtailing the exercise of market power and/or reducing SMP. The
NETA variable is statistically significant at the 99% level (p-value of 0.0092) but it does not
support the Ofgem contention that NETA caused prices to fall because the coefficient is both
large and positive. Instead, it is consistent with a rise of &3.89/MWh in day-ahead prices if
NETA had been implemented alone, in the absence of other mitigating factors. To be
2o The results published in this paper were generated using Eviews 4.0 software. As a crosscheck for accuracy and consistency the analysis was repeated using StatPro software that produced identical results to two decimal places.
33
consistent with a price reduction, the NETA coefficient should have been negative, or if it
had made no difference to prices, it should have been close to zero and/or eliminated from the
final regression model by virtue of a low statistical significance.
In contrast, the COALCCI variable is statistically significant at the 97% level (p-value of
0.0234) and also has a magnitude and direction that are consistent with a fall in day-ahead
prices brought about by divestments. Over the entire period 1 April 1990 to March 2002 the
Coal CCI fell by 3856 which, when multiplied by the COALCCI coefficient value, represents
a price fall of 27.14. The COALCONR coefficient is also statistically significant and shows
that there was a reciprocal (negative) relationship between the tonnage of UK coal that
generators contracted for and day-ahead prices. This result supports the previously cited
academic research conclusions, and the empirical observation that vesting contracts made the
market more competitive with Pool prices rising as contract volumes decreased. As the total
annual tonnage of UK coal contracted for over the life of the Pool and NETA fell by around
53 million tonnes the COALCONR coefficient is consistent with a rise of 28.80MWh in the
absence of other mitigating factors.
The CAP coefficient is negative and shows that the Pool price cap caused a temporary
reduction in SMP of around 21.94 that was reversed immediately after it was lifted. The lack
of statistical significance in the COALMC coefficient is not surprising given that Pool prices
were consistently well above short-run marginal generation costs. Indeed, the absence of
COALMC as a significant variable is consistent with the exercise of generator market power
in an imperfectly competitive market. If a fully competitive market had been operating,
especially over the life of the Pool, then COALMC Coefficient should theoretically have been
close to 1 because any change in marginal coal-fired generation costs would have
instantaneously been reflected in a change in electricity prices of an equal magnitude.
Taken together, the significance, magnitude, and direction of the COALCCI, COALCONR,
CAP, and COALMC coefficients confirm the widely held view that, the coal-fired generation
duopoly of National Power and PowerGen was able to exercise a considerable amount of
market power in the Pool. This was especially so after 3 1 March 1993 when the vesting Coal
contracts expired. Furthermore, the results suggest that without the ameliorating effects of the
Pool price cap, followed by the multiple rounds of plant divestment from 1996 onwards, the
34
two firms would have continued to exercise significant market power and raise SMP
significantly above the levels that were actually observed.
The CCGTCCI, OILCCI, and OILMC variables are more difficult to explain in the context of
the exercise of market power by mid-merit coal-fired generators. Since CCGT normally ran
as baseload plant, and oil plant marginal costs rose to such an extent that it quickly became so
uneconomic that much of it was closed or mothballed after 1993/94, neither plant type played
a significant role in price setting under the Pool. However, the result suggests that entry by
CCGT plant, and the convergence of coal and gas marginal generation costs in the later years
of the Pool, may have had a more significant impact on prices than has previously been
acknowledged. For oil plant, the CCI remained almost constant throughout the life of the
Pool but fell during the first year of NETA’s operation when one plant was sold to a new
entrant in 2001. The inherent flexibility of oil-fired plant has made it an important factor in
the operation of the BM once NETA was introduced and the statistically significant OILCCI
supports the empirical observation that oil plants displaced inflexible coal plant at the margin
in both the day-ahead market, and BM, under NETA. This may have been encouraged by
NGC because of the way in which it changed its contracting strategy for reserve capacity that
allowed oil-fired plant to cover their fixed costs associated with ramping up their boilers to
operating temperaturesz1. The unexpected reinstatement of partially mothballed oil plant also
effectively increased total available generation capacity, especially on a day-ahead and on-
the-day basis, which may have put hrther downward pressure on prices in 2001/02.
As the pricing of long-term gas supply contracts signed by new entrant P P s were partly
indexed to oil prices, the negative OILMC coefficient probably reflects the cross-price
elasticity between coal and gas. Long-term take-or-pay gas contracts signed by operators of
the early CCGT plants were indexed to oil, and still are across much of Europe. Given the
physical pipeline interconnection of the UK and European gas markets via the Bacton-
Zeebruge pipeline across the North Sea, this tends to lead to a rise in both forward and day-
ahead UK gas prices as oil prices rise. CCGT operators tend to curtail electricity generation
whenever it is more profitable to sell their contracted gas supplies rather than convert it to
electricity. This reduces the availability of CCGT capacity, especially in winter peak demand
periods, with the shortfall generally taken up by marginal coal plant capacity. This has
That is the cost of fuel burnt in the initial heating stage before electricity can be produced.
35
especially been the case since short-run marginal generation costs for CCGT and coal
converged in 1997/98.
An earlier version of these results (Bower 2002) prompted significant feedback from the
industry and other sources. The consensus view was that RPD was not the functional
equivalent of SMP, but that it had in reality replaced PPP. The ensuing discussion gave rise to
three observations:
i. excluding the impact of other factors, the industry expected RPD to be higher than
SMP as generators sought to offset the loss of Capacity Payments, and the imposition
of an additional transmission charge2’, after NETA was introduced;
ii. the duopoly generators were able to exercise market power in the Pool by
interchangeably choosing to either increase SMP or the Capacity Payment component
of PPP;
if the introduction of NETA had no impact, and excluding the impact of other factors,
then the level of RPD under NETA should not have been significantly different from
the level of PPP under the Pool.
iii.
The SMPRPD regression model, summarised in Table 2, does appear to support the first
observation that, excluding the impact of other factors, RPD was higher than SMP because
generators were attempting to offset the loss of Capacity Payments and the increased
transmission cost, when NETA was introduced. Subtracting the E1 .OO/MWh extra
transmission cost from the &3.89/MWh increase indicated by the NETA coefficient suggests
generators would have raised RPD by &2.89/MWh to compensate for the loss of Capacity
Payments, had it not been for other mitigating factors. Given the standard error of
&1.47/MWh on the NETA coefficient, and that mean Capacity Payment was &1.99/MWh
over the life of the Pool, the result therefore appears to support the industry view. In the
absence of other mitigating factors, generators would have rationally responded to NETA by
maintaining RPD at least at a level that offset the loss of Capacity Payments, and the
introduction of the additional transmission charge. However, even if this interpretation is not
correct, the result still does not support the Ofgem contention that NETA caused day-ahead
prices to fall. At best, the result suggests that changing the market mechanism from Pool to
NETA had no impact on the revenues that generators received in the day-ahead market.
’’ Under the Pool consumers paid the cost of transmission from the ‘station gate’ to the ‘gnd supply point’ where the electricity is converted to a lower voltage for consumption but under NETA generators pay 45% of this cost.
36
Though a coherent and consistent explanation of the results can be presented for the
SMPRPD regression model, its robustness is somewhat weakened by the Durbin Watson
statistic that indicates marginally significant serial correlation between the regression
residuals. In addition, the statistically significant constant term in the model indicates an
arithmetically increasing long-term trend in day-ahead prices, of the order of &1.38/MWh per
annum. This cannot easily be explained by reference to economic theory, or empirical
observation, and may have arisen because the constant term is acting as a proxy for other
dependent variables not included in the initial run, or variables that were subsequently
excluded. To address these issues, and to test the industry hypothesis observation that PPP
and RPD would have remained approximately equal if NETA had had no effect, an
alternative regression model was developed but with PPPRPD replacing SMPRPD.
3.5. PPP-RPD Results
Analysing the PPPRPD series in exactly the same way as for the SMPRPD series described
above produced the results summarised in Table 3.
Table 3: PPPRPD regression output
Variable Coefficient Std. Error t-Statistic D-value
CCGTCCI COALCCI COALCONR GASMORAT OlLCCl TEMPHEAT C
0.0014 0.0005 2.8566 0.0050 0.0018 0.001 1 1.731 3 0.0859
-0.0003 0.0001 -2.3563 0.0200 3.4807 1.3090 2.6592 0.0089
0.001 3 2.7613 0.0066 0.0036 0.6348 0.3528 1.7995 0.0744 0.9804 0.8037 1.2198 0.2248
R-squared 0.2082 Mean dependent var -0.0213 Adjusted R-squared 0.1 702 S.D. dependent var 6.5336 S.E. of regression 5.951 5 Akaike info criterion 6.4568 Sum squared resid 4427.61 19 Schwarz criterion 6.6096 Log likelihood -41 9.1 456 F-statistic 5.4794 Durbin-Watson stat 1.6710 Prob(F-statistic) 0.0000
Not surprisingly, the COALCCI, CCGTCCI, and OILCCI coefficients, identified as
significant in the SMPRPD model, remain in the PPPRPD model. This confirms the
importance of concentration in the ownership of plant capacity as a key driver of day-ahead
prices. However, the most significant change from the SMPRPD regression model is that the
37
NETA variable is no longer statistically significant. This indicates that in the absence of other
factors, RPD under NETA would have been at a similar level to PPP under the Pool. This
observation appears to confirm the industry observation that under NETA generators would
have raised RPD above the level of SMP to compensate for the loss of Capacity Payments.
The inclusion of the TEMPHEAT as a new coefficient in the PPPRPD model, not in the
SMPRPD model, is consistent with the operation of the LOLP calculation that increased the
Capacity Payment component of PPP as plant margin fell. The direction and magnitude of the
coefficient reflects the short-term impact of unusually cold weather on electricity demand for
heating in winter and suggests that PPP rose by approximately &0.63/MWh for each 1 degree
Celsius that mean monthly temperatures fell below the normal winter level. However, the
impact is relatively small as the unusually warm weather in the months December-February
of 2001/02 appears to have only contributed a &0.60/MWh reduction in RPD below the PPP
of the previous year.
The GASMORAT coefficient indicates that the gas moratorium allowed generators to raise
PPP by &3.48/MWh. Although it was only a temporary intervention, given that the
moratorium was imposed shortly after coal and gas marginal generation costs had converged,
this result supports the Ofgem view that the threat of new entry by CCGT helped to curtail
the exercise of market power by the duopoly generators. The coal-fired duopoly appears to
have responded rationally to the initial DTI announcement that it intended to leave the
moratorium in place for an extended period by raising prices. The fact that GASMORAT
affected PPP, but not SMP, also indicates that the price impact came through an increase in
Capacity Payments. This again confirms the industry observation that the coal-fired duopoly
exercised market power by interchangeably raising bid prices to increase SMP and or
withdrawing generation capacity to raise Capacity Payments.
Overall the PPPRPD result appears to be more robust than the SMPRPD result because all of
the significant variables included in the model, as well as their magnitude and direction, are
consistent with economic theory and empirical observation. In addition, the Durbin Watson
statistic now indicates that the serial correlation in the residuals of the SMPRPD model has
been eliminated. Finally, the unexplained arithmetic upward trend, previously indicated by
the statistically significant constant term in the SMPRPD model, does not appear in the
38
PPPRPD model. To further confirm the robustness of the conclusions from the PPPRPD
model, a model of the CAPACITY variable was also developed as summarised in Table 4.
Table 4: CAPACITY regression output
Va r ia b I e Coefficient Std. Error t-statistic P-value
CAP 3.1956 0.9041 3.5346 0.0006 0.0002 CCGTMC -1.4596 0.3862 -3.7798
COALCCI -0.0020 0.0008 -2.6530 0.0090 PMARGIN -0.3887 0.1391 -2.7948 0.0060 RETA 2.2270 1.0084 2.2085 0.0290 TEMP H EAT 0.4124 0.2420 1.7044 0.0908 C -0.7144 0.4369 -1.6351 0.1045
R-squared 0.2773 Mean dependent var -0.0046 Adjusted R-squared 0.2427 S.D. dependent var 4.6240 S.E. of regression 4.0241 Akaike info criterion 5.6741 Sum squared resid 2024.1 745 Schwarz criterion 5.8269 Log likelihood -367.4875 F-statistic 7.9954 Durbin-Watson stat 1.8788 Prob(F-statistic) 0.0000
The inclusion of TEMPHEAT and PMARGIN coefficients in this CAPACITY model is
consistent with the response of Capacity Payments to cold weather, and changes in plant
margin, respectively, and also confirms the hypothesis that the mid-merit duopoly could
respond to a fall in SMP by withdrawing capacity to increase Capacity Payments. In this
respect the CAP coefficient indicates Capacity Payments increased by around &3.20/MWh
when the Pool price cap was imposed and that this therefore more than offset the
&1.79/MWh reduction indicated by the CAP coefficient in the SMPRPD model. This explains
why CAP was not significant in the PPPRPD model, and is also consistent with the empirical
observation that overall Pool prices did not fall under the Pool price cap.
The positive RETA coefficient suggests that, rather than reducing prices, the regulatory
scrutiny that the review brought simply caused generators to change bidding strategies in
order to increase Capacity Payments. Likewise, the negative COALCCI and CCGTCCI
coefficient also indicates that as SMP fell due to increased competition caused by plant
divestment, and new entry by CCGT burning lower priced gas, fiom 1996/97 onwards the
duopoly generators responded by withdrawing marginal capacity to produce an offsetting rise
in Capacity Payments.
39
3.6. PSP-NETA Results
The analysis presented so far focuses on the revenues earned by generators, in the form of
SMP, PPP, and RPD, but NETA was motivated by concerns over rising prices to consumers
not the revenues of generators. It is therefore still possible that although NETA may not have
affected SMP or PPP it could have had some effect on the final day-ahead wholesale price
being paid by suppliers and ultimately passed on to consumers. To complete the analysis, a
PSPNETA regression model was developed, as described previously for SMPRPD and
PPPRPD, with the results presented in Table 5.
The PSPNETA model is less rich than the PPPRPD model, as it contains fewer variables.
However, it echoes the previous conclusions, and confirms that NETA played no significant
role in reducing prices paid by suppliers. Not surprisingly, the CCGTCCI, COALCONR and
GASMORAT variables remain from the PPPRPD model. However, CCGTMC comes in with
COALCCI, OILCCI, and TEMPHEAT variables excluded. Though this model explains much
less of the variability in the underlying data than the previous models it is still robust in the
sense that the variables that are included are statistically significant, the Durbin Watson
statistic indicates that there is no serial correlation in the residuals, and the constant plays no
significant role as an unexplained variable.
Table 5: PSPNETA regression output
Variable Coefficient Std. Error t-Statistic p-value
CCGTCCI CCGTMC COALCONR GASMORAT C
0.0014 0.0005 2.6202 0.0099 -1.2507 0.6691 -1.8692 0.0639 -0.0003 0.0001 -2.4237 0.01 68 3.9247 1.4803 2.6512 0.0090
-0.4645 0.7198 -0.6453 0.51 99
R-squared 0.1 569 Mean dependent var -0.0567 Adjusted R-squared 0.1303 S.D. dependent var 7.2628 S.E. of regression 6.7732 Akaike info criterion 6.701 0 Sum squared resid 5826.2367 Schwarz criterion 6,8102 Log likelihood -437.2633 F-statistic 5.9065 Durbin-Watson stat 1.7059 Prob(F-statistic) 0.0002
To test the PSPNETA results further a model of the UPLIFTBM variable was also developed.
This contains many more variables than the PSPNETA model, as indicated in Table 6. Again,
40
NETA appears to be an insignificant factor in explaining changes in UPLIFTBM. However,
as for the CAPACITY model, it appears that the reduction in SMP that arose as a result of the
imposition of the Pool price cap, and the intense scrutiny surrounding RETA, was to some
extent offset by an increase in revenues from Uplift. Since Uplift partly covered the increased
cost associated with generators rescheduling plant on-the-day this result also supports the
hypothesis that the mid-merit duopoly were able to exercise market power by manipulating
their bids to exploit the complex Pool rules through the strategic withdrawal of capacity.
These results not only provide evidence that this so-called ‘gaming’ behaviour took place but
that it influenced every component of the final Pool price, not just SMP.
The significance and direction of the CCGTCCI, CCGTMC, and COALCONR are consistent
with the previous discussions, though the COALMC and EASTLEASE variables appear for
the first time as statistically significant variables. The EASTLEASE coefficient indicates that
Uplift fell by &0.71/MWh once the plant leases arising from the first round of mandatory
investment were cancelled in favour of an outright sale. Although it is not immediately
obvious how the cancellation of Eastern leases, or falling coal prices, could have affected
Uplift the relatively small price response combined with the lack of an EASTLEASE variable
in any of the previous models, suggests that the mid-merit duopoly were not able to exercise
significant incremental market power through the terms of the leases.
Table 6: UPLIFTBM regression output ~~ ~
Variable Coefficient Std. Error t-Statistic p-value
CAP CCGTCCI CCGTMC COALCONR COALMC EAST LEAS E OlLCCl RETA
0.41 65 0.0002
-0.4875 -0.0001 0.21 19 0.71 15
-0.001 1 0.6079
0.21 13 0.0001 0.1022 0.0000 0.0863 0.2620 0.0003 0.2624
1.9706 2.0768
-4.7698 -4.6166 2.4544 2.71 53
2.3 166 -3.9829
0.0510 0.0399 0.0000 0.0000 0.0 155 0.0076 0.0001 0.0222
C -0.3313 0.1005 -3.2965 0.0013
R-squared 0.3249 Mean dependent var -0.0352 Adjusted R-squared 0.2810 S.D. dependent var 1.0231 S.E. of regression 0.8675 Akaike info criterion 2.6194 Sum squared resid 92.5735 Schwarz criterion 2.8160 Log likelihood -1 63.8831 F-statistic 7.3995 Durbin-Watson stat 1.5763 Prob(F-statistic) 0.0000
41
As discussed previously, marginal coal-fired generation costs were not reflected in Pool
prices because the mid-merit duopoly were able to exercise market power. The COALMC
variable here therefore probably only reflects the impact of falling coal prices on BM costs
under NETA. Although flexible coal plant fi-equently participated in setting very high SBP in
peak hours, well above short-run marginal costs, coal plants generally only ran to meet
contractual commitments, or provide reserve capacity, during peak hours. Empirical
observation of SSP and SBP suggests that generators were willing to bid this coal capacity
into the BM during off-peak hours at the short-run marginal cost of fuel in order to avoid the
fixed cost of shutting down and then reheating their boilers for the following day. This may
explain why coal marginal costs are being reflected in UPLIFTBM but not elsewhere in this
analysis.
3.7. Results Summary
Overall, the results confirm that the introduction of NETA alone, in the absence of other
regulatory interventions, would not have resulted in a statistically significant reduction in
wholesale prices. However, the RETA inquiry that preceded it did appear to be a significant
variable in both CAPACITY and UPLIF'TBM models though with a positive sign which
probably reflects the fact that intense regulatory scrutiny during the RETA investigation
caused generators to change their bidding behaviour to reduce SMP, but offset this by
simultaneously changing their bidding strategies to increase Capacity Payments, and Uplift,
in ways that were more difficult for Ofgem to detect.
Despite significant effort, attempts to collect and analyse forward prices on the same basis as
the day-ahead prices proved impossible because forward market liquidity was so poor before
1996/97. However, the fact that the NETA coefficient was positive in the SMPRPD model,
and not statistically insignificant in all other cases, casts doubt on the Ofgem contention that
the announcement that the Pool would be replaced cast a forward shadow on prices well
before NETA was actually implemented on Go Live date.
It is possible to draw this conclusion by observing the impact that NETA had on day-ahead
prices. If the expectation of NETA really had caused prices to fall for long dated forward
contracts that were due to mature after the NETA's expected Go Live date, as Ofgem
suggested, then as that date approached there should have been a sequential cascade of price
42
reductions in shorter dated forward contracts maturing after NETA Go Live as well. Taking
this process to a logical conclusion, a fall in the day-ahead price should have been the final
event in the sequential forward price cascade. However, this would only have happened at the
point of NETA’s implementation, and not before, because day-ahead prices are forward
contracts with a one-day maturity. There is no evidence in any of the analysis presented
above that NETA caused a fall in mean day-ahead prices in the month after NETA Go Live
date, as compared to the month before. Given this observation, the only logical conclusion is
that the expectation of NETA’s implementation could not have caused year-ahead, month-
ahead, week-ahead, or even day-ahead, forward prices to fall.
The real reason that day-ahead wholesale prices fell, well before NETA was actually
implemented, is revealed by the other significant variables that repeatedly appeared
throughout the regression models presented above. The principal causes of falling prices in
the last year of the Pool, and first year of NETA, were reductions in the concentration of
ownership in coal-fired generation capacity, combined with new entry by CCGT that caused
overcapacity, and an increase in plant margin.
Empirical observation of the behaviour of day-ahead prices, forward prices, as well as
contracting by generators, is also consistent with this conclusion. A graphical presentation of
one-year forward EFA prices from 1999 to 2002 in Newbery (2002) provides qualitative
support for the argument that plant divestment caused forward prices to fall. His analysis
shows that EFA prices fell by &S.OO/MWh (around 20%) in justtwo months during
February-March 2000. This price fall was immediately preceded by the contractual
completion on the second round of mandatory coal plant sales by PowerGen (Fiddlers
FerryBeny Bridge in July 1999), National Power (Drax in November 1999), as well as the
first voluntary sale of a coal plant by National Power (Eggborough in January 2000). As
McClaine (2002) notes, once the new owners had taken formal control of these plants, they
began to operate quite differently from when they were under the ownership of National
Power and PowerGen. It appears that in an attempt to recoup the purchase cost of these
plants, and cover the substantial financing costs, the new owners increased their output in
early 2000 as compared to the same period in 1999. At around the same time, it was
rumoured that a trading company, Dynegy had also sold a significant quantity of forward
contracts short, hoping to buy them back at lower prices later. The combination of these
factors inevitably had an immediate impact on day-ahead prices because it effectively
43
increased the capacity available to the market at the same time as new CCGT plants were also
being commissioned.. In addition, A E S , Mission Energy, and British Energy almost
simultaneously began to sell large quantities of forward contracts, in an attempt to lock-in
future profit margins before the end of the 1999/00 financial year-end. This forward
contracting also assumed an increase in the level of output from these plants as compared to
their previous owners. Simultaneously, operators of newly commissioned CCGT plant would
have been attempting to achieve the same objective, and the sudden increase in the amount of
forward contracts on offer seems the most likely reason why forward prices fell so sharply in
February-March 2000.
In general, the analysis shows that the impact of the other regulatory interventions, apart from
RETA and NETA, was mixed. In the case of the gas moratorium, prices increased, even
though the DTI that introduced it was simultaneously voicing concerns about the high level
of Pool prices. In the case of the Pool price cap it reduced SMP as intended but the results of
this analysis reveal that it was relatively easy for National Power and PowerGen to offset the
loss by increasing Capacity Payments and Uplift. MALC apparently had no effect on
modifying generator behaviour or day-ahead prices.
In principle the removal of Capacity Payments under NETA should have reduced at least one
potential mechanism for gaming the Pool, and in theory should have resulted in an increase in
efficiency and a fall in day-ahead prices. However, these results show that if NETA had been
introduced without the forced plant divestments that took place before Go Live date then
even this potential benefit could easily have been offset by the mid-merit generation duopoly
by a corresponding increase in RPD.
44
4. REGULATORY IMPLICATIONS
Over the period 1990/91-2001/02 analysed in this paper, the two regulatory interventions that
had the most impact were the imposition of vesting contracts for coal that kept prices low in
the early years of the Pool, and the forced divestment of coal plant that reduced prices in later
years and in the first year of NETA. The rapid reduction in the quantity of UK coal
contracted for after March 1993 revealed the true extent to which the National Power and
PowerGen duopoly were able to wield market power. It was only restructuring of the
ownership of coal capacity through forced plant divestment in 1996, and again in 1999,
combined with the unforeseen impact of the ‘dash for gas’ on overcapacity, that effectively
offset this effect and reduced prices back to the level they had been during 1990/91,
In the remainder of the paper, the regulatory implications of the findings are discussed. In
particular, a cost benefit analysis of the main regulatory interventions is attempted, and
conclusions drawn about the impact on prices if NETA is extended to Scotland and other
countries where Pool-based electricity wholesale markets still operate.
4.1. Cost Benefit Analysis
The analysis presented above, shows that day-ahead prices fell in response to coal plant
divestments not to the introduction of NETA. In addition, while Ofgem’s primary focus had
been on increasing efficiency in the wholesale electricity market, compensating mechanisms
in the capital market that had been unleashed at vesting in 1990 meant that high rates of
investment in CCGT in response to high Pool prices resulted in over capacity. Efficiency in
the wider UK energy market had also increased in response to deregulation of the electricity
market and, as the Law ofOne Price predicts, the marginal cost of coal and CCGT generation
converged in 1997/98. The analysis shows that all of these factors had begun to affect both
day-ahead, and probably also forward prices, by the end of 1999/00. In other words, at least a
year before NETA Go Live date. Given the observation that the price reductions Ofgem and
the DTI were seeking had already largely taken place before NETA was introduced, and as
there was no evidence that NETA would reduce prices once implemented, a fact that is
confirmed by the analysis in this paper, then the decision to replace the Pool was an
unnecessary and costly regulatory step. Although the Pool had some technical weaknesses,
for example in the way that the dispatch schedule was calculated, there is no evidence that its
45
centralised auction, with uniform Pay SMP pricing, was not economically inferior to the
decentralised market with Pay Bid pricing of NETA.
It could be argued that attempting to apportion the price fall to one or other regulatory action
is an irrelevant analytical exercise because the crucial economic issue is that wholesale
electricity prices fell to a competitive level, and what caused it is not important. Ofgem
essentially confirms this was its view in a single paragraph in the summary of its report on
the first year of NETA:
Before NETA was introduced the expectation was that the new trading arrangements, together with the more competitive generation market, more demand-side influence on price setting in the newly emerging markets, and lower generation input costs, offered the prospect of reductions in wholesale prices. These price reductions have occurred. (Ofgem 2002)
In essence, this argument suggests that any number of simultaneous regulatory interventions
can be justified providing the overriding objective is ultimately achieved. However, this
ignores the impact of regulatory cost and risk arising from the industry-wide systems
development and implementation that was required to operate NETA. The conclusions of the
Competition Commission inquiry into MALC effectively criticised the multi-layered
approach to regulation, that Ofgem and the DTI had adopted in the late 199Os, by refusing to
back the imposition of another regulatory intervention without first seeing the impact of
industry restructuring, and the introduction of NETA. The results of the analysis above
support the Competition Commission conclusion because it shows that the wholesale market
did become substantially more competitive as a direct result of the second round of
mandatory plant divestment that Ofgem was simultaneously implementing alongside the
MALC consultation, and NETA system development process, during 1999/00.
Ofgem’s own estimate of the direct cost of implementing NETA, published in October 1999,
suggests a figure between &136m and &146m per annum for a five-year period, and a further
&30m per annum thereafter for ongoing system modifications. In total this amounts to an
estimated direct cost of &1000m, though inevitable cost overruns have since inflated this
figure. However, this estimate does not take into account any of the associated management
costs borne by the industry, or the operational risk associated with the transition from the
Pool to a real-time trading system under NETA.
46
The cost of implementing the computer systems necessary to support NETA, and the fact that
Go Live date was later than planned, indicates the complexity of the task of introducing this
entirely new electricity trading system, especially the real-time BM. Although the NETA
transition was achieved without any system disruption, this was not a guaranteed outcome,
and the impact of a total collapse of the UK electricity system on the wider economy would
have been substantial.
In contrast, the direct costs associated with the mandatory plant divestments are considerably
lower than those for NETA. Typically, the banking, legal, and advisory costs, on transactions
of this size are around 3% of the total value. Since the first round in 1996 raised &2150m, and
&3 175 million for the second round in 1999, this amounts to a total regulatory cost imposed
on the industry by mandatory plant divestment of the order of E1 60m.
In addition to lower regulatory costs, the operational risk imposed on the transmission system
by incrementally transferring the ownership of six power plants, spread over a period of five
years, is obviously much easier to manage than changing the entire system of scheduling,
dispatch, and real-time balancing, on a single day. Since the employment contracts of the
personnel who operate power plants usually transfer to the new owner, along with the
generation assets, the day-to-day management know-how embedded in a particular plant’s
operation does not change significantly. Even if subsequent changes to the management
practices of a power plant cause an increase in unplanned outages this can easily be managed
within the normal grid operational process. It is therefore difficult to see how plant
divestment could have caused any significant increase in operational risk to the system as
compared with that on NETA Go Live date.
The estimates presented above underline the fact that regulatory interventions can impose
significant costs, and potential risks, that must ultimately be borne by consumers. Moreover,
the magnitude of those can vary dramatically depending on the nature of the choices made by
the regulator. The cost and benefit of introducing NETA versus the two rounds of mandatory
plant divestment were quite different. Plant divestment cost less than 10% of the cost of
setting up and running NETA up to March 2002. Moreover, the plant divestments that took
place imposed a one-off cost that will not be repeated beyond the date of completion. In
contrast, NETA is expected to continue requiring upgrades and maintenance for the
foreseeable future. Given that NETA also appears to have produced no measurable impact on
47
wholesale day-ahead prices, while mandatory plant divestments had an immediate effect at a
fraction of the cost, it is difficult to conclude that the introduction of NETA was anything
other than a costly mistake.
4.2. System Security
By the time that NETA completed its first year of operation it began to be criticised by both
the DTI and the industry, because of the low level of prices in 2000/01. Fears began to grow
about system security as nuclear plants were no longer profitable, and a number of coal and
CCGT plant were withdrawn from production. However, as there is no evidence that NETA
caused prices to fall, it cannot logically be blamed for threatening system security.
All the evidence in this paper points to the downward move in prices, and the closure of
plants, being caused by fragmentation of the mid-merit sector and overcapacity. Calls by
generators to reintroduce Capacity Payments, to ensure sufficient marginal capacity remained
on the system to meet unexpected peaks in demand and plant outages, were therefore largely
ignored. Neither Ofgem, nor the DTI, appeared to agree with this suggestion but the threat of
the entire nuclear industry being pushed into a default on its debt, and decommissioning
obligations to the UK Government, continued to be the major focus of concern.
In summer 2002, the DTI suggested that NETA would have to be reformed in some
unspecified way to allow British Energy, along with operators of other types of inflexible or
inherently unreliable capacity, especially renewable and Combined Heat and Power (CHP)
plants, to get a fair price for their output (FT 2002a). Already alarmed by the growing UK
dependency on gas, the DTI saw nuclear energy following the same declining path as coal,
while the output from renewable and CHP capacity had already fallen due to the impact of
high SBP in the BM. As an interim measure, the DTI intervened directly to broker a deal to
save British Energy fi-om bankruptcy by granting it a short-term E410 million credit line from
public funds (FT 2002b), while a long-term solution could be considered including putting
the company into administration and/or restructuring its debts and decommissioning
liabilities.
Meanwhile, the company, along with major shareholders and bondholders, continued
negotiations with the DTI over handing control of some of the Magnox nuclear plants, still
owned by BNFL, to British Energy along with a ‘management fee’ of E100 million. A 50%
48
reduction in the cost of fuel reprocessing services, supplied by BNFL to British Energy’s
plants, was also proposed along with a cut in business rates (local property tax) of 20-30%,
which together amounted to a further potential E100 million saving. Consideration was also
given to redefining nuclear capacity as a form of renewable energy so allowing it to avoid
paying the climate change levy (carbon tax) and producing another E80 million saving. As all
of these sums would effectively involve a direct transfer of tax revenue from the public purse
to British Energy the Treasury was reluctant to agree. As at the end of September 2002, no
solution had been agreed and the DTI credit facility was increased to E650m and extended to
December 2002. Ironically, the DTI that had so vigorously pursued the objective of reducing
prices to consumers, through increased competition in the wholesale market, was now
effectively agreeing to shelter British Energy from the impact of that competition.
In practice there was no need for any intervention, and certainly not to introduce further
NETA reforms. Wholesale prices were 11% higher in the months following March 2002, as a
result of the closure of some marginal capacity, and plans to mothball some thermal plants
were cancelled or reversed. Outages on one nuclear plant that had contributed to British
Energy’s fall in profits further increased day-ahead prices during summer of 2002.
Plant margin on the England & Wales system, which is defined by NGC as percentage
surplus available capacity in excess of winter peak demand, amounted to 26% by the end of
2000/01. Although this is low by international standards, where margins of 30% are typical,
this simply reflects an efficiency gain brought about by the introduction of competitive
electricity markets. Only if the plant margin were to fall to the previous low of 15% reached
in 1996/97 would it begin to threaten system security. However, as a hrther 5000 MW of
CCGT capacity was already under construction or in final stages of commissioning by mid
year 2002 and 4600 MW of zero registered (mothballed) capacity was available to return to
the system at short notice this does not seem likely. Although construction of a further 6700
MW of CCGT that had already received final licensing consent might not go ahead if prices
remained too low to justify investment costs, NGC estimated in June 2002 (NGC 2002) that
if only currently existing transmission contracted plant, were available in 2007/08 and no
zero registered capacity plant were able to return to the system, then the plant margin would
still be 17.3%. If all planned CCGT were built then plant margin would remain above 24%
until 2008/09. If planned closures of Magnox nuclear plant were delayed and all mothballed
plant was returned to production, plant margin would be well over 30%. If British Energy
49
were to fall into bankruptcy, then shareholders would lose their investment, and managers
their jobs, although the plants could carry on operating as before but relieved of the burden of
paying interest on some or all of their debt, and/or potentially their decommissioning
liabilities. The same applies to the fossil fuel plant that was in default, and plant margin
would therefore be unaffected.
Prices were still low by historic standards and firms were rationally responding to
overcapacity without intervention. Ofgem appeared unconcerned about system security.
Indeed, the Ofgem Director General publicly opposed the DTI plans to rescue British Energy,
instead suggesting that a natural result of competition was that some capacity would close,
and firms would fail (FT 2002a). In light of the historically high plant margin, the industry
response to falling prices by closing capacity is consistent with that observed in other capital
intensive industries such as oil refining, chemical manufacturing, or computer chip
fabrication. Reintroducing Capacity Payments, and giving special concessions to British
Energy, would effectively only subsidise capacity that was uneconomic. Both the short-run,
and eventually the long-run, economic efficiency of the wholesale market would be impaired
as a result. Indeed, as the analysis above shows the potential for manipulation of the Capacity
Payments by withdrawing capacity was significant under the Pool, and reintroducing a
similar scheme under NETA might even provide generators with adverse incentives to
undertake a joint plant closure programme so as to artificially reduce plant margin to a
dangerously low level and so increase Capacity Payments.
Not only is shielding nuclear capacity from the effects of competition inconsistent with the
stated regulatory objective of the DTI and Ofgem to put consumer interests first, but retaining
plant on the system that is not economically viable will also keep prices at lower levels for
longer than they otherwise would be. This could force other plant to be closed that would
have stayed open if prices were higher. In addition, investment in new more efficient CCGT
plant would be curtailed.
4.3. BETTA in Scotland
Using the coefficients fiom the analysis above, it is possible to estimate the potential impact
of integrating Scotland into NETA. Electricity market deregulation in Scotland followed a
different path to that in England & Wales and consumer prices will remain under regulatory
50
price control until the end of 2003/04 with the level set by Ofgem by reference to prices in
England & Wales. As for England & Wales, the biggest regulatory problem that Ofgem faces
in introducing a competitive market to Scotland is the dominant vertically integrated duopoly
position of Scottish Power, and Scottish & Southern, in the mid-merit generation sector. If
Scotland had been included in NETA in 2001/02, the limited capacity, and inadequate access
rights to the England-Scotland interconnector would have exacerbated this industry
concentration effect by generators south of the border in England & Wales.
Given that Scottish Power controlled a little over 69% of the mid-merit coal capacity in
Scotland during 2001/02, and Scottish & Southern owned the remainder, this equates to a
CCI of 5700.23 This is slightly higher than the level of the Coal CCI in England & Wales in
1990/91, and compares to a mean Coal CCI of 1382 in England & Wales during 2001/02.
Multiplying this 4343 difference between the Scotland mid-merit CCI and the England &
Wales Coal CCI by the COALCCI coefficient in Table 4 reveals that if Ofgem had lifted
Scottish retail price controls, and simultaneously introduced NETA into Scotland during
2001/02, the Scottish duopoly would have been able to reap a wholesale price advantage of
&8.10/MWh over generators in England & Wales.
The estimated price differential between England & Wales and Scotland therefore shows that
Scottish Power, and Scottish & Southern, would jointly have been able to exercise market
power that could not have been curtailed by the threat of imports from south of the border.
However, assuming that the inteconnector capacity is increased to 2200MW by the year
2003/04, and that generators in England will have full access rights to that capacity thereafter,
11 firms could potentially enter the Scottish market with 200MW of marginal coal-fired
capacity in 2004/05. Further assuming that Scottish Power closed 1000 MW of its existing
coal-fired plant to offset the likely overcapacity that would arise, the CCI for the mid-merit
sector in Scotland could fall to around 1750. This is approximately 280 higher than the
average Coal CCI in England & Wales during 2001/02. On this basis, and assuming there
will be no retail price controls in Scotland after 2003/04, that produces a forecast day-ahead
wholesale price premium of around &OSO/MWh in Scotland over the level seen in England &
Wales during 2001/02.
23 Includes the recently repowered Peterhead power plant
51
The success of BETTA will therefore critically depend on guaranteed access to the Scottish
market by marginal coal-fired plant generation capacity south of the border through an
interconnector with sufficient capacity to meet a significant proportion of total Scottish
demand. If this level of access is not guaranteed, Scottish consumers will not see the 1-2%
retail price reductions forecast by Ofgem. Indeed, prices would most likely rise unless price
controls were extended beyond 2003/04.
4.4. NETA and Other Countries
In Europe the Nord Pool market covering Norway, Sweden, Finland, and Denmark employs a
similar day-ahead uniform price auction format to the Pool, as does the Netherlands, Spain,
Germany and Poland. France also introduced a market mechanism similar to Nord Pool in
November 2001. Italy, Greece, and Austria are in the process of implementing Pool-based
electricity markets. Elsewhere, several US states such as New York, New England,
Pennsylvania-New Jersey-Maryland, as well as Alberta in Canada, employ a pool-based
system. California’s PX market also did until trading was permanently halted by the 1999-
2000 electricity price crises. In light of the experience in England & Wales there is a real
danger that regulators in countries with pool-based systems will observe the fall in wholesale
prices after NETA’s introduction, along with the DTUOfgem conclusions, and conclude that
replacing their own Pool with a NETA style market will reduce prices in their country. This
is particularly likely to happen in countries such as Spain, and the Netherlands, where prices
are well above the European average. In practice both of these countries suffer from similar
industry concentration issues as England & Wales in that there is a concentrated oligopoly of
generators in the mid-merit sector and heavily constrained cross-border transmission capacity
that prevents sufficient imports entering the market from elsewhere.
The results of the analysis presented in this paper show that it is the distribution of plant
between firms, as well as the absolute number of firms, in the generation sector that has the
most important impact on the exercise of market power in the wholesale market - not
whether a Pool or bilateral market mechanism is implemented.
4.5. Conclusion
The analysis in this paper has allowed the relative impact of the full range of regulatory
interventions that Ofgem implemented throughout the life of the Pool, culminating in the
52
implementation of NETA, to be compared for the first time. It shows that Ofgem was right to
put a cap on Pool prices in 1994 and to insist that the duopoly of National Power and
PowerGen be broken by divestment of plant. However, the results also show that the decision
to fragment the mid-merit into only three, rather than five or more firms in 1996 was a missed
opportunity that Ofgem took a further three years to rectify. The results also show that Ofgem
was right to press for the removal of the gas moratorium, and the DTI was right to continue
allowing more low-cost foreign coal to be imported, even though it displaced indigenous
deep-mined UK coal. This increased the overall efficiency of the UK energy market by
inducing inter-fuel competition between coal and gas and thereby contributed to a further
reduction in electricity prices above and beyond what industry restructuring alone would have
brought. The introduction of NETA was unnecessary, and a waste of resources, because it did
not curtail market power or lower prices. In this light, the Ofgem decision to delay the
introduction of BETTA to Scotland in 2001/02, and to continue price controls instead, was
correct given the existing duopoly industry structure.
From a regulatory standpoint, perhaps the most important finding in this paper is that a broad
policy failure occurred when Ofgem and the DTI jointly adopted a multi-layered approach to
regulation in 1996/97. This ultimately resulted in NETA being implemented without any real
proof that it would reduce prices, and insufficient attention being given to the ongoing impact
of other regulatory interventions that had already been introduced. Ironically, the results of
the analysis in this paper confirm that Ofgem’s initial scepticism over the likely beneficial
impact of ‘Pay Bid’ versus ‘Pay SMP’ market mechanisms, as expressed in 1994, was
entirely correct. NETA did not induce a price fall, and experience has shown that the BM
favours generators with larger plant portfolios containing flexible plant.
53
GENERAL REFERENCES
Bower, J. (2002) “NETA is no BETTA than the Pool” Power UK, Vol. 99, pp 35-40.
Bower, J., Bunn, D. (2000) “Model-Based Comparisons of Pool and Bilateral Markets for Electricity” Energy JournaIVol. 21, No. 3, pp 1-29.
Bunn, D. W., Larsen, E. (1992) “Sensitivity of Reserve Margin t Factor Influencing Investment Behaviour in the Electricity Market of England & Wales” Energy Policy, May, pp 420-429.
Central Electricity Generating Board (1989) CEGB Statistical Yearbook 1988/89
Competition Commission (2001) AES and British Energy: A report on references made under section 12 of the Electricity Act 1989. CC No. 453. HMSO ~vww.conipetition-commission.org.u~reports/453elec.hhn#full
Currie, D. (2002) “The new electricity trading arrangements in England and Wales: a review” and accompanying Chairman’s comments” by McCarthy, C. in Utility Regulation and Competition Policy (Edited by Robinson, C.) Institute of Economic Affairs and London Business School. Published by Edward Elgar Northampton, MA, USA. ISBN 1 84064 850 3
Department of Trade and Industry (1990-2002a) Digest of United Kingdom Energy Statistics. HMSO. ISBN 0 1 1 5 15486 8 www.dti.~ov.uk/ener~v/inform/dukes/dukes2002/~dex.sli~
Department of Trade and Industry (1990-2002b) Energy Trends: A Statistical Bulletin. HMSO ISSN 03081222 www.dti.gov.uk/enerpy/infodenergv trends/index.shtml
Department of Trade and Industry (1998) Conclusions of the review of energy sources forpower generation and Government response to 4th and 5th reports of the Trade and Industry Committee. HMSO Cm 4071. ISBN 0 10 140712 2
Department of Trade and Industry (2000) Energy Paper 68: Energy projections for the UK : energy use and eneray-related emissions of carbon dioxide in the UK, 2000-2020. HMSO ISBN 01 15154965. mv.dti.gov.uk/energv/infonn/energy proiections/index.shtml.
Department of Trade and Industry (2001) Press release: New electricity market goes live: Wholesale prices down 30%. 27 March 2001 h~:l/www.nds.coi.gov.ukicoi/coi~ress,nsf/~ti/
Electricity Association (199 1-2001) Electricity Industry Review (superseding UK Electricity). The Electricity Association, 30 Millbank, London SW 1P 4RD. http://www.electricity.org.ukJservices fi.html
Electricity Pool of England and Wales (1994a) An Introduction to the Pool Rules: Issue 2.00, The England and Wales Electricity Pool 338 Euston Road, loth Floor, Regent’s Place, London, NWl 3BP.
Electricity Pool of England and Wales (1994b) A Users Guide to the Pool Rules: Issue 2.00, The England and Wales Electricity Pool 338 Euston Road, lo* Floor, Regent’s Place, London, NW1 3BP.
Electricity Pool of England and Wales (1998) Pooling and Settlement Agreement for the Electricity Industry in En land and Wales: Schedule 9: The Pool Rules, The England and Wales Electricity Pool 338 Euston Road, 10 Floor, Regent’s Place, London, NW1 3BP. t E
Electricity Pool of England and Wales (1990-2002) Statistical Digest (various issues), The England and Wales Electricity Pool 338 Euston Road, lo* Floor, Regent’s Place, London, NW1 3BP.
Financial Times (1999) “Wholesale electricity prices to fall by at least 13%”. In: Financial Times, 4 August 1999, Financial Times Limited, London. www.€t.com/
Financial Times (2002a) “Electricity pricing faces reform” (plus other accompanying articles and comment). In Financial Times, 26 August 2002, Financial Times Limited, London. www.ft.com/
54
Financial Times (2002b) “Ministers agree to rescue package for British Energy ”(plus other accompanying articles and comment in same issue). In Financial Times, 718 September 2002, Financial Times Limited, London. www.fi.com/
Green, R. J. (1996) “Increasing Competition in the British Electricity Spot Market”, The Journal of Industrial Economics, Vol. XLIV, No. 2, pp. 205-216.
Green R., J. (1999) “Draining the Pool: The Reform of Electricity Trading in England & Wales”, Energy Policy, Vol. 27, No. 9, pp. 515.
Green, R., and Newbery D. M. (1992) “Competition in the British Electricity Spot Market”, Journal of Political Economy Vol. 100, No 5, pp 929-953.
McCarthy (2001) Letterfrom Callum McCarthy to Energy Minister, Brian Wilson MP. 3 I August 2001. httu :llwww. ofgem. ~ov.uMdocs200 1 /letter brian wilson.pdf
Maclaine, D. (2000) “Supply: Customers see large price falls” and “Pool Prices: Prices fall by 25% in the Pool”. Power UK, Vol. 72 (February 2000), pp 13-14 and pp 16-17.
Maclaine, D. (2001a) “Will NETA be Worth it?” Power UK, Vol. 84 (February 2001), pp 13-16.
Maclaine D. (2001b) “Neta: The first three months” Power UK Vol. 88 (June 2001), pp 9-12.
Maclaine D. (2002) “One Year on - has NETA been a success?” Power UK Vol. 88 (June 2001), pp 9-12.
National Grid Company PIC (1991-2002) Seven Year Statement (various issues).
Newbery, D. (1995) “Power Markets and Market Power”, Energy Journal Vol. 16 No 3,39-66.
Newbery D. (1997) “Lectures on Regulation Series VII: Pool Reform and Competition in Electricity”, The Institute of Economic Affairs, 2 Lord North Street, London, SWlP 3LB
Newbery, D. (2001) “‘British electricity restructuring: from the Pool to NETA” paper presented to the MIT Energy and Environmental Policy Workshop held on 6 December 2001 (see especially page 17). htt~:/lw7vw.econ.cam.ac.uWdae/people/newbe~lMIT200 1R-slides.pdf
Newbery, D. (2002) “’England’s Experience with NETA” paper presented at the conference International Experience with Energy Liberalization: Lessons for Europe held in Oviedo, Spain on 5 July 2002 (see especially page 3 1). http:llwww.econ.cam.ac.Llk/electricitv/news/oviedo/newberv.pdf
Parker, M. J. (2000) Thatcherism and the Fall of Coal, Oxford University Press for the Oxford Institute for Energy Studies. ISBN 0-19-730025-1.
Patrick R. H., Wolak, F. A. (1997) “Estimating the Customer-Led Demand for Electricity Under Real-Time Market Prices” (Preliminary Draft), available from http://www.stanford.edd-wolaW
Power UK (1994-2002) Power Station Tracker (various issues). Published by Platts a division of McGraw Hill Companies Inc. www.platts.com/
Wolfram, C. D. (1998). “Strategic Bidding in a Multi-unit Auction: An Empirical Analysis of Bids to Supply Electricity in England and Wales”. NBER Working Paper. http://.cvww.economics.harvard.edd-cwoIfram/paperslauctions.pdf
55
OFGEM REFERENCES
Ofgem (1990-2002) Office of Gas and Electricity Markets, 9 Millbank, London SWlP 3GE
Pool Price Inquiries and Price Car,
The following documents relating to the regular inquiries into price levels, and interventions such as the imposition of a price cap, can be downloaded at l i t tp : / /u?vw.of~em.~ov.ukl~~~bl ic /ado~loads .h~i~ool
Report on Poolprice inquiry. December 1991
Review of Poolprices. December 1992.
Pool price statement. July 1993.
Consultation on Pool reform. March 1994.
Report on trading outside the Pool. July 1994.
Pool prices and the Undertakings on pricing given by National Power and PowerGen. 26 Januaiy 1995 Generators' Pool price Undertakings 1994/95. June 1995.
Text of letter from DGES to National Power and PowerGen concerning the undertakings on sale or disposal ofplant. 27 July 1995.
Statement from the Director General of Electricity Supply on generators' price undertakings. 12 December 1995.
Report on Pool price increases in winter 1997/98. June 1998.
Independent Assessor's report on plant closure. April 1998.
Pool Price: a consultation paper by Offer. Februaiy 1999.
Pool Price: decision document. 4 May 1999.
Rises in pool prices in July. Ofgem consultation. 19 July 1999.
Rises in pool prices in July: a decision document. October 1999.
Review of Electricity Trading Arrangements ( E T A )
The key RETA documents can be downloaded from: uw.ofr.em.rrov.uklelarcWaback.htm, further subsidiary documents, including inquiry terms of reference, conference presentations, and h r d party submissions to the inquiry are available at: httu:l/www.ofrrem.~ov.ukltmblic/adownloads.htm#retabm
Review of electricity trading arrangements: a consultation paper. 5 November 1997
Review of electricity trading arrangements: Background paper 1: electricity trading arrangements in England and Wales. February 1998.
Review of electricity trading arrangements: Background paper 2: electricity trading arrangements in other countries. February 1998.
Review of electricity trading arrangements: working paper on trading inside and outside the pool. 24 March 1998.
56
Review of electricity trading arrangements: a possible common model for trading arrangements: a discussion note by Offer. 1 May 1998.
Review of electricity trading arrangements: interim conclusions. June 1998.
Review of electricity trading arrangements: proposals. August 1998.
Review of electricity trading arrangements: framework document. November 1998.
New Electricitv Trading Arrangements (NETA
The key NETA documents can be downloaded at: www.ofzem.gov.uWelarchianetadocs.htm. Many more techmcal papers relating to the systems development can be downloaded at: http://www.ofgern.gov.uWelarch/reta contents.htm
The new electricity trading arrangements: Volumes 1 and 2. July 1999.
The new electricity trading arrangements: Volume 2, Ju/uly 1999.
The new electricity trading arrangements: a draft specification for the balancing mechanism and imbalance settlement. J u b 1999.
The new electricity trading arrangements: Ofgem/DTI conclusions document. October 1999.
An overview of the new electricity trading arrangements VI. 0: A high-level explanation of the new electricity trading arrangements (NETA). 31 May 2000.
The new electricity trading arrangements: Ofgem/DTI further conclusions. August 2000.
The review of thejrst year of NETA: A review document Volumes 1 and 2. 24 July 2002.
British Electricitv Trading Arrangements (BETTA)
Key documents relating to interim adrmnistrative wholesale pricing arrangements in Scotland after implementation of NETA, as well as plans to include transmission pricing in NETA and eventually extend it to Scotland can be downloaded at: www,ofgem.aov.ukJproiectstbetta index.htm.
Transmission Access and Losses under NETA: A Consultation Document. May 2001.
Scottish administered wholesale pricing arrangements: A review of the present arrangements and consultation on prices to apply from 1 April 2002. 13 December 2001.
The Development of British Electricity Trading and Transmission Arrangements (BETTA): A consultation paper. 12 December 2001.
Application for a reservation of pre-upgrade capacity on the Scotland-England interconnector to Scottish Power Generation Ltd: Consultation Document. 2.5 October 2001.
Scotland-England Interconnector: Access & Charging Principles to apply in the transmission area of SP Transmission Limited for the period to 31 March 2002: A Consultation Document. 25 October 2001.
Interim Administered Access Arrangements on Scotland-England Interconnector: Final proposals. 6 March 2001.
Scotland-England Interconnector: Access Criteria to 31 March 2004 Summary of responses and next steps. 21 December 2001.
57
Decision by the Authority on an application for a reservation of pre-upgrade capacity on the Scotland- England interconnector to Scottish Power Generation Ltd Decision document. 25 Jantiaiy 2002.
ScotlardEngland Interconnector: Access & Charging Principles to 31 March 2002 and Access Principles to 31 March 2004. 21 December 2001.
Scottish administered pricing arrangements from 1 April 2002 Ofgem proposals document. 25 January 2002.
Transmission access and losses under NETA: Revised proposals. 26 February 2002.
The Development of British Electricity Trading and Transmission Arrangements (BETTA): Ofgem/DTI report on consultation and next steps (plus appendices) 20 May 2002.
BETTA Seminar Presentation Slides. 20 June 2002
Market Abuse Licence Clause (MALCJ
Key documents relating to the introduction of MALC and the Ofgem response to the Competition Commission decision can be downloaded at: http://www.ofgern. pov.uk/public/mbfiame.htm
Introduction of the market abuse condition into the licences of certain generators. Ofgem's initial submission to the Competition Commission: 4 May 2000.
The market abuse licence condition for generators. A decision document. 14 April 2000.
Ofgem's second submission to the Competition Commission: 22 June 2000.
Ofgem's investigation of Edison First Power under the market abuse licence condition. Initialfindings. July 2000.
The Market Abuse Licence Condition under NETA: Guidelines: A consultation document: 29 September 2000.
Press Releases
Ofgem press releases commenting on the effectiveness of RETMETA can be downloaded at www. ofgem. gov.uk/public/press releases .htm or littp://www.ofgem. gov .uk/public/pressframe .htm
Press notice: Competition in the Electricity Market. December 1998.
Press Notice: Pool Prices Must Come Down. January 1999.
Press Notice: New electricity trading arrangements Go-Live 27 March 2001.
Press Notice: Scotland needs more electricity competition to prevent it falling further behind England and Wales. 9 April 2001.
Press Notice: NETA One Year On: 27 March 2002.
Press Notice: NETA and Prices: The Facts. 6 September 2002.
58
OXFORD INSTITUTE FOR ENERGY STUDIES 57 WOODSTOCK ROAD, OXFORD OX2 6FA ENGLAND
TELEPHONE (01865) 311377 FAX (01865) 310527
E-mail: [email protected] http://www.oxfordenergy.org